Evan Lee
> [VISUAL: Text overlay "Evan Lee, Head of Partnerships | Motion"]
I talk to more top 1% creative strategists than anybody.
A grid of video call participants with the "THUMBSTOP LIVE" logo]
I stream with them on Thumbstop Live, and I have access to the top minds in the game through Motion,
Logos of brands appear around Evan: vuori, AG1, TRUE CLASSIC, RIDGE, grüns, ILIA, HEXCLAD, BYLT]
including our customers like Vuori, AG1, True Classic, Ridge, Grüns, Ilia Beauty, HexClad, Bylt, and hundreds more. And this is their process:
Slide titled "The Creative Strategy Flywheel" with a diagram showing 7 steps: 01 Research, 02 Ideation, 03 Briefing, 04 Content Creation, 05 Evaluation, 06 Launch, 07 Creative Analysis]
The Creative Strategy Flywheel. Every great performance creative team runs some version of this process. But this year, man, the tsunami of AI slapped us all around.
Evan is suddenly wearing a wetsuit and sitting next to a surfboard]
And the best strategists rode that change like a wave, reading the turns, adapting their skills.
Three pillars appear with icons and text: "DATA" (bar chart), "CREATIVITY" (brain), "AI" (robot head)]
They realized that now, creative strategy has three parts: data, creativity, and knowing AI inside and out.
Dollar signs fall around Evan]
I'm going to run you through three parts of the flywheel that the top 1% of creative strategists have superpowered with AI to generate real dollars.
Slide with text "1 Creative Analysis"]
Number one: Creative Analysis. A year ago, naming conventions were everything and mandatory to begin pulling performance trends. This was a manual process. You have to find an ad with spend, watch the video, and hopefully uncover a ton of iteration opportunities. But today, the best teams are focused on two things.
Slide titled "Meta Andromeda" with the Meta logo and a diagram showing "Ads creative", "Personalized ad retrieval", and "Ad candidates"]
Because of Andromeda, Meta's powerful AI retrieval system, iterations matter way less. Creative diversification is now everything. Using AI to categorize and recognize patterns way faster than any human could. On the creative diversification side, the best teams are shifting to ad families.
Slide titled "A VISUAL EXAMPLE OF THE SPECTRUM OF DIVERSIFICATION" showing four mobile phone screens with different ad variations: Similar Message, Same Style; Differentiated Message, Same Style; Differentiated Message AND Differentiated Style]
Take a top value proposition and explore diverse formats, creators, and visual styles to ensure you access fresh audiences.
Slide titled "VALUE PROP: HYBRID NON STICK PANS" showing three vertical video ads for HexClad with different hooks: "COOK FAST AND CLEAN FAST", "COOK LIKE A PRO", "CRISPY EDGE ON YOUR MEAT"]
On the AI categorization side of things, I know what you're thinking: where the heck do I even start? You need to start tracking what you're testing and decide what you want to learn. And trust me, if your naming conventions are a bit wonky, you want to try and find a way to clean those up.
Screenshot of the Motion platform showing a dropdown menu for "AI tags" with options like Asset type, Visual format, Messaging angle, Hook tactic, Seasonality, Offer type]
At Motion, we believe there are eight elements that every DTC brand should be tracking in their naming conventions.
Screenshot of the Motion platform showing a list of ads with various AI tags applied (e.g., Skit, Greenscreen, Podcast, Text Message)]
And speaking of cleaning up naming conventions, our new AI tagging feature can literally automate all of this for you
Evan is sitting on the couch, but the background is replaced with a digital, matrix-like overlay of numbers and graphs]
and give you superhuman pattern recognition powers. Our team of creative strategists developed prompts that read and label your creatives.
Screenshot of a Motion report titled "Winning Patterns" showing a bar chart grouped by "Visual format". A cursor clicks on the "Group by" dropdown and selects "Visual format" from a list of AI tags]
Jump into Motion's comparative reports and AI tags to uncover what formats, creators, and angles are working so you know where to implement some additional creative diversity.
Slide with text "2. Research" and subtitle "Training AI with the best references"]
Number two: Research. Training AI with the best references. By the end of this section, you'll learn how to prompt like a pro and generate great references. I get it, it's cool, you talk to ChatGPT every day. But I want to give you the skills to become a top 1% prompter. This is the highest leverage thing you can do to improve your skills in the era of AI.
A thumbnail for a video titled "HOW TO DO CREATIVE STRATEGY IN 2025 Featuring Alysha Boehm" appears next to Evan]
Remember our internal creative strategists who automated naming conventions for you? I've seen them go super deep into prompting this year. Here are the four steps to crafting the best prompts that they've learned.
A mock chat interface with the prompt: "You are an expert creative strategist with decades of experience scaling meta ads to drive revenue for multi-million dollar brands."]
Start by telling the LLM who it is and what it specializes in. You need to walk it through the steps you do, just like you would a coworker. This helps frame its tone, point of view, and the type of logic it should use. Treat the technology with respect. You're basically speaking to a PhD-level supermind, and a smart person needs context.
The chat interface updates with the next part of the prompt: "You've been provided with the following: - Creative examples from our competitors - Content about our new product - A detailed description of the primary customer persona"]
Next, clearly outline the data or information that the LLM can reference in this task. If you're giving it multiple resources, use a clean, bulleted list. Here's an example: You've been provided with the following: Creative examples from our competitor, context about our new product, a detailed description of the primary customer persona. This step ensures the LLM doesn't make up its own information and understands what real data it should ground its analysis in.
The chat interface updates with: "Your task is to identify the most common pain points targeted across our competitor ads and explain those themes that might be resonating based on the provided customer persona and brand context."]
After setting up the role and inputs, tell the LLM exactly what it's being asked to do. Your task is to identify the most common pain points targeted across our competitor ads and explain those themes that might be resonating based on the provided customer persona and brand context. This helps the LLM stay locked in on what matters, especially when the data it's working with is broad or complex.
The chat interface updates with: "Here are your instructions: Step 1: Review the summaries for each competitor ad. Step 2: Identify new value props being explored in recent ads. Step 3: Write a short explanation of how we could apply these value props to our brand based on the product and customer persona details provided."]
Step four is to give the LLM its instructions, walking it through the task. Break this into simple, numbered steps to keep the LLM focused and minimize the risk of skipping over important details. Here are your instructions: Review the summaries for each competitor ad. Identify new value props being explored in the recent ads. Write a short explanation of how we could apply these value props to our brand based on the product and customer persona details provided. The more you structure its thinking, the better your results will be. And there you have it, you're ready to prompt like a pro. Now, we want to ensure you have the proper reference points to inform your prompt. And the one final area of research that I've been geeking out about is organic social. To understand how you're going to fit into the feed and then stand out, you need to deepen your knowledge of organic social. Here's something I wish I had learned earlier: the more inputs, the better your ideas. Most people have way too narrow of a creative vocabulary. This is the immersion phase. Drink from a fire hose and just absorb.
Screenshot of the Motion platform showing a "Brand intel" page with competitor brands like HexClad and MERT. It then shows a detailed view of HexClad's ads, including a "Media Mix" breakdown and a grid of active creatives]
You can start by using Motion to look at your competitor ads and study their media mix, and save any ads to boards that you can share with your team.
A mobile phone screen showing a TikTok feed with various marketing and creative strategy videos. Text overlay: "Create a burner social account for inspo"]
But don't stop there. Build a burner social profile and begin engaging with accounts and content your ideal customer would. The goal is to get a lay of the land and understand what your customer sees.
A split-screen video call with three people. Text overlays highlight their words: "CREATIVE STRATEGIST", "IS ONE OF THE MOST", "ROLES", "TODAY", "BRANDS", "ARE ACTUALLY", "USING AI", "AD CREATIVE"]
The top 1% of creative strategists don't just live on paid, they also go deep into organic social trends.
A split-screen video call with Evan Lee and Alexa Kilroy. The Thumbstop Live logo is visible]
Alexa Kilroy, Head of Marketing at Bridge, told me something similar on a recent Thumbstop Live. There's no secret to organic social. You doomscroll, and that's how you get new ideas. Alexa is consistently studying culture phenomena and looking into how she can replicate it into the marketing context.
A Google Calendar interface. An event is being created titled "Scroll TikTok" from 12:00pm to 12:15pm, set to repeat daily]
So open your calendar right now, create a 15-minute daily time block called "Scroll TikTok", and doomscroll TikTok. You'll increase your creative vocabulary, and the ideas will find you. Once you find something interesting, what should you do with it? What I've started to do is invest more in building my reference system.
A Notion document titled "Reference System - Make a Copy". It contains sections for "ACCOUNTS", "TOOLS", "Ad Hooks - Organic Inspo", and "Customer Problem Sets". It then shows examples of TikTok profiles linked in the document (cakesbody, golinutrition, tartecosmetics)]
Begin by building a reference doc in Figma or Notion. I like to then break it out by creative themes. I've also posted a link to my template so you can copy it in the companion resource.
Slide with text "3. Ideation & Content Creation"]
Number three: Ideation and Content Creation. Ideas have always been cheap, but AI makes them damn near worthless. We have too many ideas, so you need to test before you invest.
Two text boxes appear next to Evan: "1. Script Writing" and "2. Generating MVP creative assets for briefing"]
These are the top two ways to use AI in the ideation and creation phase. Number one: script writing. Number two: generating MVP creative assets for briefing. The old approach was weeks of back and forth with creators before getting to a final output. The new approach is generating the best script possible with the context you have on your consumer, and turning that into a minimum viable asset with AI so your creators and editors can see exactly what you want. You can start by generating customer segment reports with ChatGPT using all the prompt tips from above. Next, extract a ton of potential ad angles from the output. Then to finish, build three to five creatives per angle and deploy them in a dynamic creative set.
A purple slide with a list of pro tips: "Always use a reference image or reference video to get better outputs from gen AI tools", "For product shots, keep the product in your hands", "Don't waste time trying to make gen AI fix small mistakes", "Don't treat these gen AI tools like a button you push"]
Here are some pro tips: Always use a reference image or reference video to get better outputs when working with Gen AI tools. For product images, keep the product in your hands. Sometimes the best performing ads come from quick shots you take yourself. Don't waste time trying to make Gen AI fix small mistakes, like weird hands. Again, like prompting, don't treat these Gen AI tools like a button you push. You still need to interact with them, apply creative judgment, and learn how to improve the output rather than giving up.
A Venn diagram appears. First, two circles: "Data" and "Creative", overlapping to form "Creative Strategy". Then, a third circle "AI" is added, creating a trifecta]
Creative strategy began at the intersection of data and creative. But now, it's the trifecta: data, creative, and a deep knowledge of AI. And that makes you freaking unstoppable. Really go deep into these areas we covered. If you do, I promise you, the best creative strategist can now do the work of four to five people. And guess what? You can earn more respect across growth and creative teams, and a big payday ahead for your career.
The Motion logo appears on a light purple background
