The ad expert behind Thesis, LMNT, and Marsmen told us how he does it.
Share
The ad expert behind Thesis, LMNT, and Marsmen just told us exactly how he does it.
Adam Greenfeld has written ads that have seen over a million dollars in spend. He's spent the last decade building and scaling some of the fastest-growing brands in wellness. He joined the FOC community to break down exactly how direct response works — and why the best ads are almost never the prettiest ones.
From our FOC monthly community call · Featuring Adam Greenfeld, VP of Ecommerce at LMNT & Founder of Fridge Creative Studio
Adam Greenfeld grew up in upstate New York with Tony Robbins playing in the background. He spent his 20s in the front office of the Brooklyn Nets, helped move the team from New Jersey to Brooklyn, then on a Wednesday when he was 30, quit his job and bought a one-way ticket to Thailand. He walked around Southeast Asia for six months and came back knowing he needed to build something that mattered.
What followed was a decade of building inside the ecommerce world: a nootropic brand called Go that failed after a year, then Thesis, which went from $200K to $35 million in four years, then a 25-person creative studio called Fridge where he made direct response ads for brands like Marsmen, Armra, Hexclad, and Ultra. Now he's VP of Ecommerce at LMNT, a $485 million business built to last 30 to 40 years.
He brought all of it to this call.
The one cognitive bias behind hundreds of millions in sales
Adam opened with the framework that has driven more revenue for him than anything else. He calls it problem evangelism, and it works like this: the person or entity that best articulates your customer's problem automatically becomes the most qualified to solve it.
It's a subconscious trigger. When someone scrolls past an ad that describes their life with uncomfortable accuracy, something clicks. They don't consciously think "this brand understands me." They just feel it. And that feeling translates directly to clicks and purchases.
The key is going three levels deep on the problem. Most brands stop at level one.
|
Level 1 — Primary "I can't focus." Easy to articulate. Almost nobody pays money to solve this version of the problem. |
Level 2 — Secondary "I stay late at work." Getting warmer. Now we're into consequences. People are starting to consider spending money here. |
Level 3 — Tertiary "I missed my kid's softball game." This is where people pay $80 a month without hesitation. Family, money, and relationships. That's the real purchase motivation. |
The ads that reliably hit seven figures in spend are the ones that get to level three. The tertiary problems that consistently convert: anything tied to money and income, anything tied to family and close relationships, anything tied to long-term health. And yes, anything that makes someone feel more attractive.
The ad that put $10 million on the board
Adam's most effective ad ever for Thesis had nothing to do with nootropics, cognitive performance, or brain optimization. It was a story about a 15-year-old kid with ADHD and a strained relationship with his mom.
He sat in his office. His intern recorded it. He told the story of how his undiagnosed ADHD meant he just wanted to watch TV while his mom wanted him to do his homework, and how those years between 14 and 18 created real distance between them. Then, in his late 30s, he connected that back to what Thesis had done for his ability to focus and show up.
He sold a nootropic to adults by telling a story about a teenager's relationship with his mother. That ad ran for 24 months in various forms and, accounting for LTV, is responsible for around $10 million in revenue.
What's actually working right now
Adam broke down the current paid social landscape with specifics. A few things worth saving:
|
Account structure 50% static, 50% video A year ago video dominated. That's shifted. Big accounts are now split evenly. If you're overindexing on video, rebalance. AI tools work much better for statics, so lean there if you're building out AI creative. |
Creative that converts Make it ugly on purpose A Post-it note on a product. A whiteboard with four bullet points. A screenshot of a notes app. A Reddit post. These have converted just as well as $15,000 productions. Stop making it pretty before you know what message works. |
|
Dark mode tip Black background wins If you're running notes app or Twitter screenshot ads, test both light and dark mode. In 38 out of 40 tests, dark mode outperformed. The black block disrupts the scroll in a way white doesn't. |
Research shortcut TikTok is your swipe file Type your category into TikTok search and study what's going viral. It shows you exactly how people are talking about your product's problem space. Great artists steal — figure out what's working and adapt it. |
|
Readability rule Write for a third grader Even the smartest people are 40 IQ points lighter when scrolling Instagram at 9pm. One technical term per video max. Put your script into Claude and ask how to say it to a kindergartener. They ran the same creative at 10th, 5th, and kindergarten reading levels. The simpler version often won. |
AI creative Feed it customer comments The best AI copy comes from projects trained on real customer reviews and ad comments. AI video creators rarely convert well — maybe 1 in 300 or 400 ads. But AI for copy variations and static visuals is genuinely useful if the inputs are right. |
The future: AEO and why Reddit matters more than you think
Adam closed with something most brands aren't paying attention to yet. He calls it AEO — Answer Engine Optimization. It's SEO, but for the LLMs. When someone asks ChatGPT or Claude "best electrolyte," the model converts that into dozens of different questions and scrapes the internet for answers. The brands whose websites are structured to be easily readable by those models win that search.
Step one is converting your backend Shopify data into a JSON-LD file. Beyond that, he's building agents that query what questions the LLMs are asking daily, then using another agent to write blog content answering those questions with scientific credibility.
And Reddit: up to 12% of the research behind some LLM responses is sourced from Reddit. If you're not paying attention to what's being said about your category on Reddit, you're missing a significant slice of how AI search will surface or bury your brand.
What we took away
- —The entity that best articulates the problem is automatically perceived as most qualified to solve it. Lead with the problem, always.
- —Go three levels deep. The surface problem is never what people actually pay money to solve.
- —Ugly ads work. Figure out your message before you invest in production.
- —50/50 static to video. Rebalance your account if you're not there.
- —Test dark mode on screenshot ads. 38 out of 40 times it wins.
- —Simplify your copy to a third-grade reading level. One technical term per video, maximum.
- —Don't try to create new habits. Find people who already have the problem and offer them a better solution.
- —Start optimizing for AEO now. Convert your Shopify backend to JSON-LD and start paying attention to Reddit.
- —Know your numbers. AI makes it easier than ever. No excuses.
One thing that stood out across the whole conversation: Adam is as honest about his failures as he is about his wins. Go failed. Thesis outgrew him. He stepped away when he wasn't the right person to keep running it. That self-awareness, knowing what you're good at and when to hand off, is as much a part of his edge as anything he knows about ads.
This recap is from our FOC monthly community call. Adam Greenfeld is VP of Ecommerce at LMNT and founder of Fridge Creative Studio. You can find LMNT at drinklmnt.com.