How AI Transformed Creative Strategy for Winning Facebook Ads (The 2025 Evolution)

Alysha Boehm: Hi, I'm Alicia, and I'm here to talk about how to do creative strategy in 2025, how AI has changed everything and what the best teams are doing about it. I'm even wearing my creative strategy hat today. I was also telling Melissa I've been spending way too much time with my product team lately and I'm very excited to put my creative strategy hat back on today. So, let me take you through everything,

Alysha Boehm: starting with a bit of my timeline. I love to talk about this because I come from the creative side of things, not the media buying side of things, which I feel like is a little bit more rare. I just don't hear about it very often. So I love to share where I came from. So back in 2022, I worked at Pela. They also do Lomi. They sell compostable phone cases and a kitchen countertop composter. And I was hired as their graphic designer and one of my responsibilities was to make their paid ads. And I was introduced to Motion, and one thing led to another. I got this PDF in my inbox called Becoming a Creative Strategist and I realized, oh my God, this is what I want to do when I grow up. So I loved making ads so much. I moved over to Kulin. I was still a designer and I was an ads designer, just making ads all day, living the dream. But I made it pretty clear I wanted to be a creative strategist right away. Kind of elbowed my way into becoming a creative strategy lead and absolutely loved it. Helped build the team and build some systems over at Kulin. And now I'm here at Motion and I am something called a context engineer, which basically means I'm really good at AI and prompting and systemizing creative strategy. Um, so that's a little bit about me.

Alysha Boehm: And just because my ads are my little babies and and I like to share my creative sometimes. This ad on the left is the one that started it all back in 2022. I lovingly refer to it as my festival ad, and so does the team here at Motion. Evan knows all about it. It's the first ad that I really followed performance for when I opened up Motion and it just like is near and dear to my heart. And then the ad on the right is one for Bouy. I still make UGC for Bouy. They were one of my clients at Kulin. And this one was a true story. I had a really great experience with their product and I just loved making it and it ended up being a top performer. So these two things were just like really, really important to me and I thought I would share them just for funsies.

Alysha Boehm: But that's enough about me. Let's talk about you.

Alysha Boehm: So tell me if this sounds familiar. If you're anything like the teams that I talk to every day, and I talk to all kinds of teams every single day, big and small, you want winning ads, you want scalable ads, and you want to unlock new audiences on Meta. This is probably the most common thing that I hear from all different teams is I want to unlock new audiences on Meta.

Alysha Boehm: So you go on LinkedIn and X to get some advice from the experts,

Alysha Boehm: and this is the kind of thing that you see. I love all of these experts, by the way. They all share so much great value, but it feels like a lot. It's so overwhelming. And you can get onto LinkedIn and you're you're seeing everything from make ads with AI, try Gumloop, MCPs, ad families, N&N, every team is doing this, don't fall behind. And it just feels like a lot of pressure.

Alysha Boehm: So then you go ahead and you try to

Alysha Boehm: try to follow their advice, and it feels like feels like this. It feels like this is just not working for me. What am I missing and am I falling behind?

Alysha Boehm: Because your expectation and the expectation that I genuinely think is being set up on LinkedIn and X right now is that ChatGPT and VO3 will generate ready to launch ad creative. Claude is a top 1% copywriter and can script better than you. AI workflows are easy to build and will generate big returns.

Alysha Boehm: But the reality, and the reality that I feel every single day, and let me know if you can relate, is that the model in the image that you generated with chat still has three knees. Claude recommended seven hooks that all sound like an infomercial, and your custom workflow is outputting broken nonsense.

Alysha Boehm: So if this sounds familiar, I'm going to let you in on a little secret.

Alysha Boehm: This is a totally normal experience, and I hear this from teams every single day.

Alysha Boehm: So what you see on LinkedIn is what is what the 1% looks like. And you might not be one of the 1% and that's okay because I'm not either. Every day I talk to the other 99% of people who are just doing the work and trying to figure it out as they go. So if that's you, you're in the right place.

Alysha Boehm: Because today is all about level setting. We're going to talk about what creative strategy looked like before, what's fundamentally changed in the last six to 12 months, and what the best teams are actually doing about it today and what you can do today to level up.

Alysha Boehm: So first, let's start start talking about where we started. So a year ago, which I can't believe, September of last year, a bucket list item happened for me. I was asked to speak in the Creative Strategy Summit, and I presented a 10 minute day in the life of a creative strategist where I walked through the creative strategy flywheel and what every step in this flywheel looked like for me as a creative strategist in my day-to-day. And I still get messages on LinkedIn almost every single day from people who say, I watched this 10 minute session and it gave me so much clarity about what what my job looks like.

Alysha Boehm: So this is what I'm hoping to unlock for a few people today who just really need to root themselves back into what, what the fundamentals of this job really looks like today. So this creative strategy flywheel, if you're familiar, people talk, and I've always talked about creative strategy being the bridge, the uh the thing that bridges the gap between growth and creative teams. And the flywheel is the process that strategists follow to do exactly that.

Alysha Boehm: But tell me if you can relate. When I was a creative strategist in the day-to-day, it started with my relationship with my creative team starting and ending with writing briefs and my relationship with the growth team starting and ending with reading the data that they shared with me. And I truly existed in the middle and actually kind of felt very siloed.

Alysha Boehm: Then it started to feel like I became more and more involved in the planning and creation of new assets because I needed to be in order to plan my creative tests. And I also felt way more involved in things with the growth team where especially for me from the creative side of things, I really had to dive deep and understand how my creative was being tested and scaled so that I could read the data effectively.

Alysha Boehm: And lately, it feels like things have merged even more. And now I feel responsible for creating assets for testing, and I feel responsible for targeting audiences with my creative. So it's not that I'm like taking over these two jobs, it just means that the overlap has become so much more significant and I've just gained so much more responsibility in my role.

Alysha Boehm: It really went from I bridged the gap to I am the bridge is how I would describe it.

Alysha Boehm: The creative strategist role has evolved and expanded a lot in the last 12 months. And this is in my opinion, the direct result of two really big shifts that happened in the space as a result of AI.

Alysha Boehm: So let me talk about these two shifts. The first one was the shift on the growth side,

Alysha Boehm: where Meta's AI and machine learning algorithms optimized for ad placement, budget allocation, and audience targeting to improve performance. Now, like I said, I come from the creative side, so please talk to your media buyer about all of the nuances about how your creative is being tested and scaled because I never feel comfortable digging deep into it. I always had a really strong relationship with my media buyer to understand what was going on in the ad account.

Alysha Boehm: But the takeaway here is that creative is the new targeting has never ever, ever been more true.

Alysha Boehm: So before, media buyers used to build these really complex targeting stacks and you could unlock a lot of growth by finding the right targeting mix. And I hear people who were in this space long before I was and long before I even knew what creative strategy was, talking about how you could carry bad creative a long way with targeting. You could you could create just about any ad and make it work on Meta. And that is quite the opposite from true today.

Alysha Boehm: So now Meta decides who sees your ad, and a lot of people actually would argue that it does a better job than humans do. And you unlock new audiences with creative, not campaigns. So you really can't media buy yourself out of a bad creative situation.

Alysha Boehm: So the two common pitfalls that I see creative strategists fall into as a result of some of these changes on the growth side is one, these are pretty obvious by the way, but but you'll see. The one is Meta optimizes for the easiest conversion. So you replicate a top performing ad, but that ad was just scooping up all of your warm leads. It wasn't unlocking new growth, which is what you really need. So you plateau because you only spoke to the people who were primed to buy and you forgot to move people through from awareness to consideration to conversion.

Alysha Boehm: And the second pitfall that I see is that Meta now favors diverse creative, really diverse creative. And people refer to this as ad families now. So you dupe a top performing ad, but the ad isn't a fresh take and doesn't introduce a new entry point. So you plateau because you're only speaking to one person in one way and you've missed a lot of opportunities to speak to different people in different ways to move more people through your funnel.

Alysha Boehm: So what this really means for you as a creative strategist is that you need to meet customers where they're at, move them from discovery to purchase, and make creative that scales efficiently. You also need to establish and uphold the systems that will break you out of a plateau and keep you out of it. Sounds intimidating. Like, no biggie.

Alysha Boehm: The next big shift was on the creative side

Alysha Boehm: with generative AI. Generative AI, no kidding, helps you generate usable creative assets like images, videos, and voiceovers in minutes with machine learning and existing data.

Alysha Boehm: Before, creative production was slow, it was expensive, and it was resource constrained. You had to hire a creator, work with a photographer or videographer, work really close with the creative team, and you had to be really selective about what ideas you tested. So for me, on the agency side, what that looked like was if the client didn't provide me with that creative asset, I was out of luck because I might have might have had a really great idea, but I couldn't produce it. But if you were in house, you would have to find a way to produce it and work really close with your creative team to make that happen. So in some cases, you had to choose between speed and volume or quality.

Alysha Boehm: But now, AI can generate usable creative assets in minutes. So asset availability isn't a bottleneck anymore. Your ideas are. And you get more shots on goal for less time and money.

Alysha Boehm: Now, there's two common pitfalls that I see here. And the first one is expecting AI to come up with great ideas right out of the box. So you ask an LLM to generate new hooks or ad concepts, but they don't compare to the ones that you come up with on your own. So you give up because you were told that AI could do your job and it didn't. But that's because you were missing some really key context that exists in your brain and that you weren't able to unload onto the LLM.

Alysha Boehm: The second pitfall, and this is especially true with creative asset generation, is expecting AI to get it right on the first try. So you type in a prompt that you found on LinkedIn, and the result needs a lot of finessing. So you give up because you expected AI to be magical. And this ad on the right is a really great example of this. This is a fake product and a fake ad, but I've used ChatGPT to generate this. And these these are the things that happen all the time is that ChatGPT will duplicate text, so you can see it says even skin tone twice. That happens all the time. Also, if you scroll through LinkedIn, this same font is on virtually every ad creative. And it never gets your product right. Now, I'm not saying that chat is the be all end all of image generation. There are tools that can do a way better job. But when you're told on LinkedIn that this is magical and this happens to you, it's kind of disappointing, right?

Alysha Boehm: But what this means is you need to prioritize your best ideas, not just the ones you have assets for. You can achieve creative volume and creative diversity without burnout. But you're not just a creative strategist anymore, you're a prompt engineer. And it's your responsibility to learn how to use these tools in the most effective way possible so that you can leverage them to make good things happen.

Alysha Boehm: So what's actually changed about the role of a creative strategist then?

Alysha Boehm: So, three years ago, I would argue that creative strategy was a bit of a buzzword. I remember being at Pela and the reality was they weren't going to hire a creative strategist anytime soon because, like Evan from Motion always said, the creative team could share the creative strategy hat. And what I said to Evan back then was, I don't want the hat, I want the job. And then I got it. But today, creative strategy is really the epicenter of paid advertising. So if there isn't isn't a creative strategist on your team, there's a lot of shared responsibility on the creative team to pick up all of this slack. And if there is a creative strategist on the team, they have a lot more responsibility than they did a year ago.

Alysha Boehm: So three things are true, which I kind of just touched on. You have more responsibility than ever for what actually happens in the ad account. You also have more tools than ever to do your job well. And the unfortunate truth, maybe not unfortunate, is that AI won't replace you, but it will accelerate good strategy and expose bad strategy. So if you're bad at your job, AI will just help you do a bad job faster. You still have to be a really good creative strategist.

Alysha Boehm: All right, so let's talk about what this looks like day-to-day. So I'm going to walk you through four parts of the creative strategy flywheel that I think have been impacted the most and how they've changed and what they look like today for creative strategists. Starting with, we're actually going to start with analysis.

Alysha Boehm: So the definition of analysis on that creative strategy flywheel is validate hypotheses and memorialize findings. If I were to rewrite this definition today, I would say it's about identifying patterns, operationalizing learnings, and guiding future tests.

Alysha Boehm: So what this looks like is tracking the components of your ads that have the most impact and enable you to track and identify performance patterns.

Alysha Boehm: So what you're probably doing today is you might already recognize these patterns in your ads. This format always works. This hook type is a top performer. This audience has never worked for us. This creator generates the most engagement. These are patterns that you know because you know your creative inside and out. But you had to look at every single ad and be really, really comfortable with them in order to find them.

Alysha Boehm: Most teams keep these insights inside of their heads because tracking them in a document or something feels really intimidating. But when I talk to teams, they speak in the same languages. They know these things about their creative. They know what works and what doesn't, what they've tried before. The best teams use naming conventions or tags to track and measure creative component performance. And the 1% plugs all of this information into a custom workflow to auto-generate insights.

Alysha Boehm: So what is a naming convention? Let's just establish that before I keep moving on. A naming convention is a system for labeling ads to help surface key insights about performance that are related to the actual ad creative. We're not talking about like hook scores or CTR or anything like that. So your old naming convention might have looked something like this, UGC Vid 125A, which tells you this is a UGC video and 125 and A probably means something, but what do they mean? I don't know. You still have to watch the ad to figure it out. A naming convention might look something like this. So now I know that this is a UGC video and it has a certain format, a messaging angle, a persona, a product, and a creator. And now I can track these things in a naming convention.

Alysha Boehm: So why would you go through the effort of using naming conventions? It's because when you apply naming conventions to your creative, that enables you to spot, filter, track, and report your insights so much easier in a tool like Motion, especially when you're working on a team.

Alysha Boehm: But the old way of doing naming conventions brought you back to spreadsheets, which if you saw the intro video, nobody likes spreadsheets. We don't like them. We're the antithesis of spreadsheets. So this felt like kind of going back in time a little bit. This spreadsheet felt really complicated. And again, when I talk to teams every single day, if they don't have a naming convention, it's because this looks and feels complicated and they just haven't gotten around to tracking it yet. They already have all those things in their head, so why would they go through this effort?

Alysha Boehm: So before this overcomplicated spreadsheet was honestly filled with a lot of ambiguous values. So I found myself tracking a lot of things that at the end of the day, I didn't feel like were actually moving the needle. They were just in the sheet or things that people talked about tracking, but like at the end of the day, they didn't mean anything to me. Also, this spreadsheet was so time consuming, difficult to maintain, and created a lot of friction because I worked at an agency and a lot of people touched the sheet and the naming conventions. And then the really unfortunate thing was that every mistake impacted my ability to track those insights effectively. So if I tagged something with the word happy, but somebody else tagged something with the word happiness, now I was tracking two different things.

Alysha Boehm: So now with AI, you can do this automatically, but you can only the key thing is that you should only track the things that move the needle. So I talked about how I tracked a bunch of different things in this naming convention. Truly, at the end of the day, if only two things really matter to you because you're on a very small brand, then you should only track those two things. Or if five things matter to you, you should track those five things, but you should just focus on the things that matter and don't let yourself get overwhelmed. Now, when it comes to using AI for this, and I'm going to talk a little bit about, about that in a second, it can take you less time, less maintenance, less frustration, less friction, and less mistakes if you're asking the LLM to walk you through this process and to help you with this process. So instead of going to the spreadsheet and thinking and manually like dropping down all of these values in the sheet, the LLM can reduce a lot of that mental load by noticing things that you might not. So it might, for example, like when I was in that sheet, I might be like, is it this ad format? Is it that ad format? If you just ask the LLM, here is my naming convention, here are the values that you need to track, it will just make that decision for you and reduce a lot of that mental load.

Alysha Boehm: So how to start is you're going to start by tracking the most impactful components of your ads, and then layer in new categories as they become important to you. And these are the four, actually these are the eight, but these are the four on the left here that I hear teams talk about the most, which are the ad format, the hook or hook type, the messaging angle, which usually touches on emotional and psychological triggers, and the audience. And the other ones that I hear, which sound a little bit more obvious would be things like product, creator, season or offer, and then visual setting. But like I said, you can tag anything that you feel like moves the needle, but I'm telling you the the four on the left are the ones that I hear the most often.

Alysha Boehm: So how to leverage AI for this task is you're going to drop your ad into Gemini with a repurposable prompt. And yes, I do have a prompt for you, but I'm not going to demo this because I don't want to waste your time showing you how I use Gemini. The prompt will be at the end of the deck and you will get it when you get this email. You're going to drop that into Gemini so that it auto generates the values that you care about and formats it for you. So it takes a lot of that effort away. I will tell you when I went through this process and named all of my ads, sometimes it would take me as much as 15 or 20 minutes, and I wish I had ChatGPT or Gemini to be able to do that for me.

Alysha Boehm: And now you do. And then what you're going to do next is you're going to apply those naming conventions to your ad names or you're going to use tags to start tracking right away.

Alysha Boehm: And if you're asking, can't I just automate this inside of Motion? The CEO told me I could tell you soon.

Alysha Boehm: So then what? Then, this is the most important part. You're going to conduct your analysis through the lens of these values and ask thoughtful and strategic questions like, why do these ad formats resonate with this audience? Why does this messaging angle work in this ad format? Why does this messaging angle not work for this product? Or you can even ask things like what ad formats am I not trying? Or should I hire this creator again? So you can see it gives you some really valuable insights.

Alysha Boehm: All right, the next part that I want to talk about is the research part of the creative strategy flywheel.

Alysha Boehm: So the definition of research in that flywheel is comment analysis, tagged social, and persona building. And if I were to rewrite that, I would rewrite it as surface brand context, customer drivers, and paid social patterns.

Alysha Boehm: So before, this step in the flywheel was for a lot of things, but it was also for internal use for training, realigning with team members, and for me, brainstorming new ad concepts and really coming back to the basics whenever I got really stuck.

Alysha Boehm: And this is still true, but now it's also for training the LLM on your brand and sharing the essential context that you need to generate good outputs. So in the example that I used like at the beginning of this deck where I just said, give me some new hooks and I said you're missing essential context,

Alysha Boehm: this is the context. So what this looks like is, and this is a screenshot from my presentation a year ago when I was at Kulin. It's everything you would need to train a junior creative strategist about. So think about all of the context that you have in your brain that helps you make creative decisions for your brand every single day and all of the snippets and insights that usually contribute to good ad creative for you. So you can see on the right here, these are the things that I would track every single day. These were like the product details and comparisons, my customer personas, press and review sites, customer reviews, competitor analysis. All of this essential information we always kept on hand, honestly just to like use in our own brains, but also to work together as a team and all share the same context. It's the same principle, but with AI, now you need all of this context documented so that you can give it to the LLM.

Alysha Boehm: So there are three types of research that you want to focus on if this is relatively new to you, or if you're like, I haven't documented anything, where do I start? The first things you're going to want to write down are your brand constraints, information about your products like features and benefits, things like your tone of voice, all of the key information about your brand. The next thing is everything about your customer that you know, pain points, emotional drivers, personas. You can also run, there are several AI tasks that will dive through your customer review CSV and give you a lot of these valuable insights. And then the other thing is information about other brands, not necessarily just your competitor, but other brands, what formats, messaging angles, and visual styles they are testing right now.

Alysha Boehm: How you stack up. So most teams keep this information in scattered sheets and briefs and Slack messages, or worse, in their heads. So I can't tell you how many times I've asked people, if you left the team tomorrow, who would have the most context? And they would just go, uh, I don't know, I would have to write it down somewhere. The best teams keep a consolidated, living, breathing document with all of this necessary context used by their team. And the 1% will feed all of this into an LLM or turn it into a custom GPT that becomes a really valuable tool for their team.

Alysha Boehm: So if you want to start today, I would document the most impactful context that you use in your work every single day. Literally every time you're analyzing an ad or planning a new creative test or conducting analysis or doing a brainstorm, just think about what context kind of pops up into your brain that helps you make those creative decisions and really makes you the expert. So these are things like brand tone and constraints, product features and benefits like I mentioned, golden nugget reviews, or even your top hooks and headlines and things that have worked and haven't worked for your brand.

Alysha Boehm: And when you're leveraging AI for this task, it's kind of ironic because you need all of this context in order to use AI effectively. But you can also use AI to make this a lot easier for you and to have all of this documentation done for you. So with one quality prompt and a CSV of reviews, ChatGPT can get you 80% of the way there. And I will also include a prompt for this in the deck that will get delivered to you. Again, I just didn't want to demo how to use ChatGPT and and use up your time that way. Uh, but you can just ask ChatGPT to help you generate this context and get you a lot of the way there. And then you're going to layer in all of the nuance, the tone, the constraints, the unique insights, and the marketing positioning that maybe ChatGPT didn't quite pick up on. The things that only you know because you're a person who works at that brand.

Alysha Boehm: And then what you're going to do next is you're going to update it every time you remember something important or you learn something new. And you're going to use it to provide context to LLMs when you're brainstorming. So when people like Alex Cooper give you incredible advice about how to use AI and he says, oh, you just need to dump all of the context in, instead of going, what context? Or this is the first step I'm going to get hung up about. You already have all of that information.

Alysha Boehm: Okay, the next step is ideation.

Alysha Boehm: So the definition for ideation on this flywheel is based on research, determine your plan of attack. And if I were to rewrite this one, I would write design and prioritize creative tests and plan for learnings at scale.

Alysha Boehm: So what this looks like is, again, every single team I've ever spoken to has some version of what I used to call my creative testing pipeline, or some people call it like a creative testing sheet or a creative roadmap. But basically, it's a sheet where they plan what they're going to test next, and they're noting dates or sprints, concept type, and whether something's an iteration or a net new concept. But if you're lucky, they might also include how many ads are going to be in those tests, the formats they're testing are, what the creators they're using are. It just depends on each brand, how much level of detail. I've seen everything from really granular detail to just like, I'm going to do a static next week.

Alysha Boehm: So before, I would see that when people were planning their next creative tests, they would really over index on metrics for very tiny little iterations. And what I mean by that is like the hook score was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing.

Alysha Boehm: So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct.

Alysha Boehm: And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach.

Alysha Boehm: So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway.

Alysha Boehm: Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space.

Alysha Boehm: So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test.

Alysha Boehm: But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are.

Alysha Boehm: So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it.

Alysha Boehm: Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct.

Alysha Boehm: Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this.

Alysha Boehm: The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing.

Alysha Boehm: So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here.

Alysha Boehm: So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators.

Alysha Boehm: But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources.

Alysha Boehm: Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% obviously have entire teams dedicated to exploring AI tools.

Alysha Boehm: So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them.

Alysha Boehm: So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips.

Alysha Boehm: With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating.

Alysha Boehm: But there are a couple things that I love to do with VO3 and that I think are very easy to to do today. One of them is also a visual metaphor, which I'll show you in a second here. One of them is a scroll stopper. I know I keep mentioning Alex Cooper, but he's a great example. He's wonderful at AI and he talks about making hooks that are really scroll stopping and then throwing those in front of some of his best ads. And then also turning statics into videos. So I don't know if you'll be able to see this video. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great.

Alysha Boehm: And then the last thing that I'm going to show you

Alysha Boehm: is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source.

Alysha Boehm: Okay, so not AI related, but the last thing I want to mention is to get the product in your hands.

Alysha Boehm: So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads.

Alysha Boehm: Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out.

Alysha Boehm: So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck.

Alysha Boehm: So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay.

Alysha Boehm: Thank you.

Melissa Rosen: Thank you, Alicia.

Melissa Rosen: Everyone,

Melissa Rosen: give Alicia lots of gifts of clapping hands.

Melissa Rosen: In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted.

Melissa Rosen: Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one.

Melissa Rosen: Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one.

Melissa Rosen: Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one.

Melissa Rosen: Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one.

Melissa Rosen: Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before. So I feel very comfortable saying this could work for you, but it's not a great strategy. And I would see a lot of people just drop in competitor ads and add inspo into their creative pipeline and call that a day. And it's really not the way to approach creative testing. So now what you want to do is start from learnings that are measurable and impactful. Those were the learnings that we talked about in analysis where we talked about tracking those really key, most impactful pieces of your creative, the ad format, the messaging, the hooks, and the audience primarily. Then you're going to build a systematic, systematic tests that the algorithm will favor. So we talked about how the meta algorithm loves ad families. So when you give it a family of ads instead of four of virtually the same ad, you're actually going to benefit from that a lot. And you're also going to make room in your flow to follow your gut instinct. And I call this a learned concept, which I'll dig into in a second. And their gut instinct usually pays off because they're tracking and paying attention to the most important things about their creative. And then the 1% hire multiple agencies to multiply this approach. So what is a learned concept? I've always called this thing a learned concept in my creative pipeline spreadsheet, whatever I wanted to call it, a road map. And what I basically positioned this as is an approach that builds off of core learnings without over indexing on metrics. It's basically something in between an iteration and a big swing. It's not really a small iteration, it's not a big swing because I learned something from it, but it's a really thoughtful, systematic test that's rooted in a really strong takeaway. Then, when we're talking about gut instinct, I don't want to discount that this is what the LLM can never do. Can't do it, can feel. So I am of the opinion that at the end of the day, when you run these ads on Meta, there is a human on the other side that's making a purchase. And the fact that you are also a human is a very valuable thing that I never want to let go of. So if you come up with a creative concept, if a creator delivers you something that's unexpected and interesting to you, and if you just feel in your gut that something might work, I think it's important to make space for that. And we're starting to hear this a lot from creators. I literally just heard a podcast from Alex Cooper the other day where he said that one of his client's top performing ads that spent like $400,000 has a hook that for all intents and purposes, he would not say is good, but it's working incredibly well. So I just want to give you permission today to be really good at your job and make space for the places where you're just like, I just feel like this is the right move and I don't ever want to lose that because I think that's going to be really valuable as we continue to move into this AI space. So what this used to look like, and by the way, these are like AI generated ads that I would not consider good. So please don't like be like, wow, she's a really crappy creative strategist. This is just an example. But what this used to look like is this would be a creative test, right? And we've all done this where you just put a different headline on the same ad and you call it a creative test. And I'm not saying that you shouldn't or can't or this could never work because I do still see some people doing it this way. Usually people who have very large, very large brands and lots and lots of budget and space to test. But if you're being more precious about your creative tests, especially, you want to do something more like this where this is rooted in, you've got this same creator and maybe she's talking about self-care, but we've got a native text selfie style headline static ad. And then we've got a UGC comment response, and then we've got a lifestyle product image with social proof. So basically you're rooting it around one or two of those components, but really setting up a really diverse creative test is the whole point. And again, not saying these are great ads, these are just examples and I honestly wish I had an extra day to make you some better ones, but here we are. So why this matters is you're going to prioritize meaningful, systematic ad concepts over small tweaks and wildcards. You're going to build an ecosystem of ads that work together to achieve diversity, volume, and scale. And you're going to get more shots on goal and the algorithm is going to reward you for it. Where you're going to start is you're just going to switch from a react to a systematic approach to your creative testing. So you're going to track those really impactful components of your ads and build your creative tests around those. You're going to build off of your learnings before your ads fatigue. This is a really important part about not being reactive. Before the ad even fatigues and you start to freak out about it, you've already learned the most important thing about it and added new variants into your creative testing pipeline. And you're going to maintain a structured system of measurement, but also never underestimate your gut instinct. Okay, lastly, we're going to talk about content creation. And this is the thing that we see the most buzz about on LinkedIn, so I'm very excited to talk about this. The definition of content creation on this flywheel is create the asset with variants. But I would swap this to generate MVP assets and accelerate ad creative into testing. So what this looks like is for me, if I'm being completely honest, there was always this unspoken step between ideation and briefing, which I kind of touched on, where you were limited by the assets that you had before you could even prioritize a creative test. And this is the part that I'm talking about here. So before, you would have to hire a creator, a photographer, or a videographer. You would go through multiple rounds of revisions and back and forth conversations, and then you would finally launch the ad weeks or months after you had this idea. And this is still going to be true for many, many things, especially working with real UGC creators. But when it comes to generating images and video, it doesn't have to be this way anymore because now you can use AI to generate a minimum viable product to test and validate your idea. And if it shows positive signals, you can always invest more into it. And then you can achieve volume and diversity with minimal cost and resources. Most teams are really struggling to implement these generative AI tools into their current systems. What I hear most is there's usually a couple of AI heroes on the team that are really, really advanced and they've explored them all. And there are a few that are really just not about it. They're just a little bit too busy and they haven't had time to catch up. And I think it's going to take a little while for the for the best teams to really align. But the best teams are encouraged to explore generative AI tools to solve for strategic, high impact use cases and there's space for it. So if you're a the lead of a creative strategy team, for example, or a growth manager, a CEO, making space for your teams to explore these so that they can move at the same pace, I think is really valuable. And then the 1% have entire teams dedicated to exploring AI tools. So where to start? I get asked this question all the time of like, okay, but like there's ChatGPT, but what else is there? What else would you recommend? And there are lots of things, but these tools, I would say are the very core ones that where I would start today. So ChatGPT for image generation, but there are some conditions, which I'll talk about in a second. VO3 for video generation. There's also conditions there. 11 labs for voiceover generation. And then I recently found this new tool called sync.so for voice cloning and lip sinking, which I love and I'll show that to you. And then beyond that, just explore as much as possible because these are the ones that you hear about the most and I would say are the most ubiquitous for last for lack of a better word. But there are lots more that you could explore that I haven't even found yet that will do a better job. You just have to like really invest the time into exploring them. So let's talk about ChatGPT first. The two things that I don't think that ChatGPT does really really well right now, other tools might, but chat doesn't, is generate text. It's always the same font, always, always, always. It duplicates text all the time. You can even see in here, it says like drawer closing or something. Like it doesn't spell things right. It's very frustrating. And also with product images, if you share your product image with ChatGPT, it just redraws it and it looks funny. So if you've seen these things happen when you're trying to generate an image from ChatGPT, you're not alone. And there's a lot of things that you can do to make it work better, but they take a lot of time. The things that I love ChatGPT image generation for are, the first one, visual metaphors, which is what this is today. So what I did was I generated a prompt for a visual metaphor for loop earplugs. And the idea here is like you're turning the volume down on, you know, like your stressful surroundings with the loop earplugs. So I love the ability that I can make this visual metaphor in this really cool image and test it in an ad without having to hire a creator or go sit at a coffee shop myself and set up a tripod and take this picture. The other thing that I like to do is make images with text and product placeholders. So you can ask it to generate the image with text and the product, but then you can bring it into Photoshop and remove those things and put the actual product image that you have and the actual text that you want into the image, but it's already taken into account of where that text and the images are. So that's where I'm at with ChatGPT. So just a couple of tips. With VO3, a lot of people are showing entire ads on LinkedIn, but when I use VO3, I could still only do eight seconds and three videos a day. So I don't know what they're doing. They must have a million emails. Um, so the other thing that I that I struggle with a little bit with VO3 is product placement. It's just like a bit frustrating. So this was a visual metaphor that I generated from a prompt where basically it came up with the idea for Ridge Wallet, if you're familiar, to put a brick in somebody's back pocket to represent like what your wallet used to feel like. And then Veo was able to generate this video. So again, instead of like hiring someone or grabbing this like very awkward video of somebody's like butt with a brick in it, I was able to generate this in VO in in like no seconds flat. And I can test this in an ad, which is great. And then the last thing that I'm going to show you is sync.io. Sync.io, first of all, don't ever use this without permission because this is scary. And I used my own ad or my own UGC clip to to demonstrate this for that reason. And I would not use this for entire videos. But what it enables you to do is clone somebody's voice and relip sync the clip. So let me show you my original clip. So that's me actually talking and recording a clip for a Bouy ad. And this is what I got with sync.io. Isn't that so crazy? I did not say that. And even though you might pick up on the nuance that this is definitely AI, when I sent this to some of my friends and my teammates without seeing this little watermark in the top right corner, they actually didn't know which one was AI and which one wasn't. So I think that this would be a really interesting tool to sub in for reshoots because one of the things when you're hiring UGC creators is you would often have to have them reshoot something, but their hair would be different or their vibe would be different or their lighting would be different. So now with their permission, you could simply just sub this in. And you can also test new ad components. So if Bouy were to ask me for permission to like lip sync my lips so that I could say a different hook, I would absolutely be okay with that with permission and probably with extra payment, obviously. Um, but I think this is going to be really interesting for UGC and solve a lot of those pain points with the time and effort that it takes to source. Okay, so not AI related, but the last thing I want to mention is to get the product in your hands. So as a creative strategist, I can't tell you how many times I achieved a winning ad. Uh, the Bouy one is a great one, a great example where I filmed the UGC myself, or I would have the ad or the product in my hand and I would like take a flat lay picture of it on my desk or outside or I would take a UGC product shot with it and then that would become a top performer because I just had a compelling thought about this creative asset that I needed. I had the product on hand, so I made it happen. So not AI related, but if you're a creative strategist and you don't have the product in your hand today, honestly, the way that I would would perceive it is that you own the ad now, you don't just own the brief. So you have the tools in your hands, whether that's AI or the product, and you should make ads with them. So make Barry Hott proud. Basically, make more ads. Okay, going to wrap it up with some takeaways. So what we covered today is that AI is at the core of everything we do now. It's happening. There's nothing we can do about it. The goal though is still to make winning ads that scale. The process, which is the creative strategy flywheel is still pretty much the same. We're just going about it a little bit differently. There's a lot more work to do, but there are also more tools at your disposal to help you do that. And the best teams are still figuring this out. So the four things that we talked about today, just to summarize, are track winning patterns instead of over indexing on metrics, document your valuable context to share with LLMs and leverage AI effectively, build the systems that prioritize ad families over siloed ideas, and then creating testable MVP creative assets to remove the creative bottleneck. So that's a bit of a summary of everything we covered today. And that is everything from me. And now I can stop talking, which will feel like a huge relief.

Melissa Rosen: Yay. Thank you, Alicia. Everyone, give Alicia lots of gifts of clapping hands. In the chat, we had some incredible feedback. Awesome. Thank you. Clap, clap, clap. Yay. Yeah, I think everybody was hanging on your every word to the point where they were like, wait, wait, wait, wait, wait. Wait, wait, what'd you say? Like go back to the slide.

Alysha Boehm: Don't worry, you will get the deck. And I put, I promise you, I put extra prompts at the end of the deck. I just didn't I didn't think it would be very useful to open ChatGPT and show you how I pasted a prompt. I promise you'll get prompts in the deck.

Melissa Rosen: Totally. I I think a lot of this people will need to re-review it with the recording, with the deck. Like, this is so much. There's so much that has changed in creative strategy and I think like, to try to put it into a 30 minute presentation is very ambitious and yet Alicia did it. So it's okay that we need to go back and review some of these some of these things. Um, but yeah, we have some time for Q and A. Um, so let's look. I'm going to pull, let's see what the most upvoted. Oh, this one I like. Let's do, let's start with this one from Maria. Can you give an example of an ad family? How would the ads differ?

Alysha Boehm: Yeah, I think I showed but it wasn't a very good one. So basically, like the difference between the creative testing that we used to do, which would be like the simplest way would be like a hook test. So you would test the same body of UGC, but you would test a different hook on it. And that's still a valid way of of creative testing, but if those were the only things that you ever tested, you would not achieve an ad family. So an ad family would be like different formats of UGC. So think about rooting yourself in, let's say there were the the four things that move the needle. The ad format, the messaging angle, the hook, and the audience that you're targeting. So hook testing is one. But when you're thinking about UGC, there's more than one ad format to UGC. There are unboxings, there are reviews, there are day in the lives, there are POVs, there are comment responses. There are, like all different kinds. Savannah Sanchez has the best test examples of different kinds of ad formats and different kinds of hooks that you can test. So if you're the whole point of an ad family is really just to look at multiple different things that could that could move the needle and just think about building a more diverse creative test. And I think there's always the fear of like, well, if I don't have like one control variable, how do I know for sure that that was the thing that moved the needle? And if I'm being fully honest, like, you never really know that anyway. So especially if you're a smaller brand and you can only run a certain amount of creative tests, like even once or twice or three times a week, you really just want to diversify as many of those components as you possibly can to move the needle along as much as you can. So an example of an ad family might be like, you have the same creator who has the same hook or a similar hook, but they're like doing an unboxing and they're addressing a certain pain point, or they're responding to a comment addressing the same pain point, or they're doing a day in the life addressing the same pain point. So instead of doing like a hook test, you're locking into one particular pain point, but you're demonstrating it in different formats in UGC if that makes sense. So you kind of notice how like a little bit less granular and a little bit more like a family of like a true diverse creative, creative being the keyword test. So hopefully that helps a lot. I wish I had given more more examples in hindsight, but yeah.

Melissa Rosen: Yeah, no, this is great. And we've already got, but someone's like, can you do a part two where you dive deeper? So, yeah, we totally can.

Alysha Boehm: I love that.

Melissa Rosen: We can work on, we can work on a part two of event. Maybe we go into like taking each step of the flywheel and having you just go crazy with examples and by steps of how to do everything. Um, I think we can make that happen. Um, let's see. Next one. I like this one. Oh, this was another one too of like just diving deeper. You mentioned like learned concepts versus iteration. Let's just talk a little bit more about what that looks like.

Alysha Boehm: Yeah, okay. So like when I talk about iteration, what I talk about is I feel like whenever anybody's introduced to Motion, and I'm actually not entirely sure if this is still true because obviously I've been using for Motion forever, but I feel like one of the biggest selling points is that you can see all of these granular little aspects of performance on an ad. But I find that some people can just like really zoom in on one thing. Like the thumb stop rate was 10% lower than usual, but our CTA was on par. So let's say swap out the hook, which is not necessarily a bad thing. It's just a bit of an outdated and very granular way of looking at your creative, especially if you're running on a smaller account and you have to be really precious about what creative you're going to launch in the account because you only have so much budget. The other thing I would see is people would test multiple versions of virtually the same ad when they were planning, or they would dupe a competitor's ad and expect it to work, which I've seen lots of experts talk about before.