Creative Diversity Beyond Volume: The 8 Ad Formats Scaling on Meta Right Now

Across 70-plus Meta ad accounts, 20 of them spending over $1 million a month, one pattern is impossible to miss. The brands that are still scaling are not running one format harder. They're running six or seven different formats at the same time, in the same account, all the time.

For a year, that wasn't necessary. In 2025, the Meta auction rewarded whoever had the most output. More volume meant more surface area, more surface area meant more winners, and "we make 500 ads a month" was a genuine competitive advantage. Agencies built their whole pitch around it.

Then AI made volume free. A teenager in his bedroom can open Claude or ChatGPT and ship 100 ad variations in 30 seconds. Any junior designer with an AI image tool can produce in an afternoon what used to take a three-person team a full day. Volume isn't a moat when everyone has it. It's table stakes.

This chapter is about what replaced it: creative diversity driven by judgment. And the eight specific formats that are working right now.

Why raw volume stopped working

When everyone uses the same tools, they get the same output. AI is trained on everything that already exists, so if you use it for the idea, you get the average of every idea that already exists. No new angles, no new framings. The whole category ends up looking like one brand made all the ads.

The consequence is that fatigue collapsed. Creative that used to take three weeks to burn out now burns out in three days. When every ad in your account looks the same, audiences tune out faster, so you get more volume and a shorter lifespan per ad. You spend more to get less, and the dashboard takes a month to tell you.

So the question "how many ads should we make this month" is the wrong question. The right question is: how many different stories are you telling in a week?

The 8 formats that are winning

The accounts spending over $1 million a month run this full mix, all at once.

Text-heavy ads. A wall of text on Instagram shouldn't work, but it does. Your brain doesn't filter it as an ad because it doesn't look like one. It looks like an email, a LinkedIn post, a long-form story. By the time you realize you're being sold to, you've already finished reading. Some brands are spending six figures on a single text ad that breaks every design rule.

Ugly ads. Crayon fonts, Post-it notes, kids' drawings. Ugly stops the scroll, and every operator knows it. One coffee brand runs an ad that looks like a child drew it: "Dad's new favorite coffee" in crayon, dad flexing next to the product on a table, no offer, no urgency. That ad doesn't need to convert on click. Its only job is to make you stop. The brands using ugly well aren't replacing their offer ads. They're stacking ugly top-of-funnel in front of polished bottom-of-funnel in the same account. One creates the stop, the other closes.

Native ads that don't look like ads. A gut supplement brand runs a photo of a woman in a hospital bed, gown on, monitors blinking behind her. The copy opens: "I had a partial bowel obstruction last April," then a wall of text telling her story. The eye reads it as a Facebook post from someone you don't know but probably should. By the time the product appears, you've already invested in the read. The best place to reach someone who hates ads is somewhere they're convinced there are none.

Highly creative ads with judgment behind them. Another coffee ad: a bikini model walking on a beach with the coffee bag where her head should be. A cappuccino mug with a face next to her. A French press with a face behind her. Sunset, palm trees, pink overlay text. Putting a coffee bag on a model's head is a human decision. The execution is AI. That's the version of AI that scales: volume of weird, distinct ideas, not volume of one safe idea.

Yapping videos. A girl getting ready and talking for seven minutes about whatever. A guy in his car on the way to Starbucks talking about his mental health. The product is in frame because it's around them, not because it's the topic. Yapper videos have crossed $1 million in spend in a single account. A founder talking to camera is the highest-converting ad in DTC right now. If your founder isn't in the account, that's where the next 10x lives.

Podcast-style videos. Two people sitting across from each other talking about a subject the brand actually cares about. The bet is length: three to five minute clips, not 15-second hooks. The product gets referenced because the brand sponsors the conversation, not because the conversation is about the product. It works because the viewer already opted into that format somewhere else, on YouTube or Spotify.

Animated videos. Cartoon characters inside a stomach explaining how a probiotic works. Animated figures pouring vitamins into an organ. Especially strong in health and supplements, because the format makes complex biology digestible while someone scrolls. Anything that has to teach a mechanism to sell (gut, sleep, focus, hormones) benefits from it.

That's seven distinct formats. The eighth is the discipline of running all of them together and rotating constantly, so no single look fatigues the account.

AI as the brain vs AI as the hands

There are two ways to use AI in advertising, and only one of them still scales.

AI as the brain, with a human as the hands. You ask AI for the idea, then execute what it gives you. The output is mass-produced, looks like every other AI ad in the feed, and Meta isn't rewarding it anymore. This is the trap most operators fall into.

AI as the hands, with a human making the call. A human decides to do something nobody else would (the coffee bag on a model's head), and AI executes it. The judgment is human, the production is AI. The brands using it this way ship a high volume of distinct, weird, deliberate ideas instead of a high volume of one safe idea.

The difference is where the judgment lives. Somebody has to decide what's worth making, what angle each format hits, and what emotional state the customer is in when they see it. AI doesn't make those calls. It produces the average of what already exists.

Why this is harder than volume

There's a clean version of this story where AI made advertising easier and you just embrace it. That version is wrong.

Running different formats means different production systems. There are no economies of scale per concept. Statics is one workflow, video is another, animation is a third, podcast-style is a fourth. Every concept is a different machine. You can't industrialize variety the way you can industrialize volume.

The brands winning aren't winning on production as a starting point. They're winning on judgment, and the volume comes after. The bar gets higher every year while everyone else races to ship the same AI ad as their competitor. Volume isn't a strategy anymore. It's a consequence of doing the work right.

The good news: you don't need to master all eight formats at once. Pick two or three that fit your product (a founder yapping video and an animated mechanism explainer for a supplement, say), run them alongside your polished offer ads, and rotate. The point is variety of story, not a specific number of ads.

The checklist

• Stop asking "how many ads this month" — ask how many different stories you're telling this week
• Run several formats in the same account at once and rotate them so no single look fatigues in three days
• Stack ugly or native top-of-funnel in front of polished bottom-of-funnel — one creates the stop, the other closes
• Get your founder on camera — a founder yapping video is the highest-converting format in DTC right now
• Use an animated explainer for any product that has to teach a mechanism to sell
• Use AI as the hands, not the brain — make the weird human call yourself, then let AI execute it
• Accept that variety has no economies of scale — each format is its own production machine, and that's the moat

Next: [Native Ads and the Advertorial Funnel — The 3-Step System That Sells to Cold Traffic →](06-native-ads-and-the-advertorial-funnel-the-3-step-system-that-sells-to-cold-traffic.md)