Why swapping hooks no longer counts as diversifying your creative, what Meta changed to make that official, and how to reset your monthly production cadence around net-new concepts.

Let's get into it →

Back in June I made a video about how marketers misunderstand creative diversity. Too many think "diversity" means creating a bunch of different versions to split test. To be fair, that was correct only a couple years ago. But now that the ad platforms rely on creative for targeting, campaigns need to be fed "unique" creative to properly scale.

Meta just agreed with me in writing.

On August 26, Meta rolled out a Creative Diversity rating, still marked "in development," that grades your ad set Low, Medium, or High based on how visually different your images and videos actually are. And the early reports are brutal. Accounts running statics, video, UGC, and carousels are still scoring Low, because every one of those assets is the same idea in a different wrapper.

Meta's own creative partnerships lead, Ashley Foster, put it plainly: "Creative diversification trumps quantity in 2026." The hook-swap era is over. Here's what replaces it.

To the algorithm, a hook swap is one ad. Only a net-new concept counts as a new one. 

Think about it like a race weekend.

Split testings hooks is like changing tire pressures on the same car. You might find a tenth, but you likely won't find a full second.

A new concept is like bringing a different car entirely. Everything about the ad is different, that it has the same goal: make a sale. 

Why Meta stopped rewarding variations. Its retrieval system, Andromeda, reads the creative itself before the auction: the visuals, the copy, the pacing, etc. Ads that look and say the same thing get grouped and treated as one candidate. Twenty hook variations on one script don't give the machine twenty options. They give it one option and nineteen copies. Change the opening line and the system still recognizes the same ad underneath.

Why it matters more now than it did in June. Placement controls are gone. Targeting is mostly automated. Meta has spent two years removing every lever you used to pull, and creative is the one it left in your hands. Then it built a scoreboard for how well you're using it. When the platform publishes a rating for something, it's telling you what it's about to reward.

What "diverse" actually means. Meta's own framing lists the dimensions: a different hook is one of them, but so is a different creator, visual style, messaging angle, offer, pain point, emotional appeal, and format. A concept is new when it changes the middle of that list, not just the first item. Same product, different reason to buy it.

The data behind the push. Meta's Performance 5 framework, which made "multiple concepts in parallel" the default, reported advertisers who adopted it saw 32 percent better conversion efficiency and 9 percent more incremental reach. The independent agency data points the same direction: near-duplicate creatives compete against each other inside the account and inflate costs.

1. They count ads instead of concepts.

The weekly report says 24 new ads went live, even when Andromeda only saw two.

2. They think format is diversity by itself. 

Static, Reel, UGC cut, carousels, etc. These do increase diversity but if they all follow one script and one proof point, they'll still score low.

3. They treat net-new creative as a quarterly event.

One big shoot, one hero concept, then three months of iterations off the footage. That worked when the machine rewarded volume. Now the concept fatigues in weeks and every iteration after that is a copy of a tired idea. The pipeline has to move at the speed the concepts burn out, which is monthly, not quarterly.

Meta stopped counting your ads. It started counting your ideas. Budget accordingly.