Almost every performance ad is one of sixteen layouts
We tagged roughly 4,500 live Meta ads against our own taxonomy of static ad layouts. Sixteen skeletons covered nearly all of them — and the craft lives in the constraints, not the ideas.
Nearly every static Meta ad we classified uses one of sixteen layouts: Before / After, Benefit List, UGC Testimonial, Us vs. Them, Headline-Led Claim, Social Proof Stack, Offer / Urgency, Problem → Solution, Feature Callout, Press / Authority, Stat-Led, Lifestyle In-Use, Screenshot Review, How It Works, Founder Note and Unboxing Flat-Lay. Each has one job, and fails when a second element competes for it.
Scroll a competitor's Meta Ad Library page for ten minutes and the first impression is variety. Hundreds of creatives, different products, different palettes, different photographers.
Tag a few thousand of them and the impression collapses. The same skeletons keep coming back. A hero shot with a stacked checkmark panel. A vertical split with a label on each side. One enormous number with a small line under it. There is a finite vocabulary here, and it is much smaller than the category pretends.
We ended up with sixteen.
What are the most common static ad layouts?
This is our own taxonomy, and we've since classified roughly 4,500 live ads from the Meta Ad Library against it. It isn't a standard anyone publishes; it's what fell out of trying to describe what we were actually seeing.
| Layout | What it does |
|---|---|
| Before / After | Side-by-side proof of transformation |
| Benefit List | Product hero beside a stacked checkmark panel |
| UGC Testimonial | Casual user photo with a quote card |
| Us vs. Them | Two-column comparison against the old way |
| Headline-Led Claim | One bold claim, minimal everything else |
| Social Proof Stack | Reviews and numbers carry the ad |
| Offer / Urgency | Deal-led banner with a deadline |
| Problem → Solution | Name the pain, then reveal the fix |
| Feature Callout | Annotated product with labeled arrows |
| Press / Authority | Endorsement or expert-framed credibility |
| Stat-Led | One striking number does the persuading |
| Lifestyle In-Use | Product alive in its natural moment |
| Screenshot Review | A phone-screen message rendered as app UI, not a photo |
| How It Works | Numbered three-step strip showing the routine |
| Founder Note | A short signed note, plain type on a quiet background |
| Unboxing Flat-Lay | Everything in the kit, top-down, with a piece count |
Note what is not on the list. There's no "beautiful product photo" layout, because a beautiful product photo isn't a layout — it's an asset that several of these consume. The skeleton is the arrangement of claim, proof and offer. Everything else is fill.
The interesting part isn't the list
Anyone can read sixteen names and nod. The list is not the useful artifact.
What's useful is the negative space. Tagging a few thousand of these teaches you far more from the failures than the successes, and the failures rhyme: each layout has exactly one job, and it dies when a second element competes for that job.
A Stat-Led ad exists to make you register one number. Put a second number beside it and you haven't made a stronger ad, you've made a chart — and nobody reads a chart in a feed. A Founder Note works because it looks unproduced; every piece of decoration you add moves it toward looking like an ad pretending to be a note, which lands worse than an honest ad. An Us vs. Them comparison carries its argument in the visual difference between the two columns, so styling them identically removes the argument while leaving the layout.
The general rule generalises further than you'd expect. A second headline, a second number, a second discount badge, a product scaled up until it rivals the claim it's supposed to support — the specific mistake varies, the shape doesn't. Whatever the layout is for, protect it, and be suspicious of anything you're adding that competes with it.
Why this matters beyond taxonomy
Three practical consequences.
Briefing gets easier. "Make me an ad" is an unanswerable prompt, for a human designer or a model. "Make me a Stat-Led ad, the number is 94%, the context line is 'of users kept it past month three'" is a brief someone can execute in one pass. Picking the skeleton first removes most of the ambiguity that makes creative rounds expensive.
Testing gets honest. Most "creative testing" is five variations of one layout with the background swapped. That tests photography, not persuasion. If you want to know whether your audience responds to proof or to price, you have to ship a Social Proof Stack against an Offer / Urgency — different skeletons, same product. Variation within a skeleton is a much smaller experiment than teams think it is.
Competitor research gets structured. The Ad Library is overwhelming because it's an undifferentiated grid. Classify by skeleton and it becomes readable: this brand runs mostly Lifestyle In-Use and has never tried a comparison; that one leans entirely on Offer / Urgency, which usually means they're buying on discount rather than on product. What a competitor doesn't run is often the more useful signal, and you can't see an absence until you have a list to check against.
The honest caveats
A taxonomy is a lens, and lenses distort.
Some ads genuinely don't fit any of the sixteen, and "none of these" has to stay an available answer or the taxonomy starts inventing matches. Carousels and video complicate the picture, since the skeleton can change between cards. And the sixteen aren't equally common: a handful account for most of what's running, and several are rare enough that seeing one is itself notable.
The bigger caveat: classification says nothing about performance. We can tell you what layout an ad uses. We cannot tell you what it spent or what it returned — for an ordinary commercial ad the Ad Library publishes no impressions, no spend, no reach and no demographics at all. Those fields exist only for political and social-issue ads, and for ads that delivered to the UK or EU. Anyone quoting performance figures for a normal brand's creative is inferring, not reading.
There's a subtler trap for anyone counting by hand. The Ad Library's default
result order appears to be impressions, high to low — not newest first.
Meta doesn't document this, but the interface writes the sort into its own URL
(sort_data[mode]=total_impressions) the moment you run a search, and the
results are visibly non-chronological. The other option, "Most recent," isn't a
strict date sort either — it groups by month and ranks by relevance inside each
group.
So a quick scroll through a competitor's ads shows you their highest-impression creative first, which is exactly the sample most likely to make one archetype look dominant. If you're eyeballing rather than tagging systematically, you're reading a ranked list and mistaking it for a representative one — and you're being handed Meta's own reach ranking without ever seeing the numbers behind it.
Treat "this brand runs a lot of X" as evidence about their strategy, not proof that X works.
The practical version
- Pick the skeleton before you write anything. It's the highest-leverage decision in the brief and it takes ten seconds.
- Give the layout one job. Whatever it is — a number, a claim, a deal — nothing else in the frame may compete with it.
- Test across skeletons, not within one. Five backgrounds is not a test.
- Audit a competitor by what's missing, not by what's abundant.
- Don't infer performance from frequency. Nobody outside the ad account knows what actually worked.
The reason the list is short is the reason it's useful. Direct response has been converging on these arrangements for decades — across print, TV and now feeds — because they map onto a small number of ways to make a stranger believe something. New formats keep arriving. The skeletons underneath them mostly don't.
Sources
- Meta Ad Library API — which fields exist, and that spend, impressions and demographics are limited to political and issue ads and to ads delivered in the UK and EU (accessed 2026-09-15)
- Meta Ad Library — a fresh search opens with
sort_data[mode]=total_impressionsin the URL (observed 2026-09-15)
Related: why one ad resized into five is a bad habit, the four Meta ad sizes that actually matter, and what the Ad Library API will and won't give you if you were hoping to do this programmatically.
Terrads builds from these sixteen layouts directly — you pick the skeleton, and each placement renders at its native size instead of being cropped down from one master.