Competitor Meta
Ad Insights

We pulled every currently-live UK ad for all nine pages, then split each account by media type to get a format on every ad and to beat the grid cap. That gave 1,330 unique ads at roughly 97 percent of what Meta reported. From those we selected 83 for the expensive work: the real creative asset downloaded, films cut into three-frame strips, and every one sorted onto a fixed checklist of 21 content types.

Selection was weighted by how long Meta has kept an ad running, because longevity is the only performance signal available from outside an account. An ad live 150 days is a proven argument. An ad live 3 days is a guess.

Each account, what it is doing, and the one thing to take from it.

Arguments counts distinct opening copy concepts. Ads per argument is the repetition rate: low means near-unique creative every time, high means one idea flooded. Offer percent is the share of copy carrying discount or free-gift language.

Sézane was captured mid-launch for a 23 August drop, so its 1-day median reflects recency, not churn. Wild Nutrition and Sézane needed a second pass after a rate-limit block; see method notes.

Median days live per content type, across the 83 creatives we sorted by hand.

Read this as a relative comparison of proven creative, not as absolute proof that one type outlasts another. Those 83 were chosen partly for longevity, so the population is not neutral. The comparisons in finding 1 are drawn from all 1,330 ads instead, which is why those are the ones stated as fact.

Eight standard arguments, each tested against all 1,272 ads carrying copy. This is the map of what is taken and what is free.

What to make next, in priority order, with the reason from the data. P1 items are open or near-open lanes with proven durability elsewhere.

Each brand's six longest-running live ads. Ordered by days live, and brands ordered by their best ad.

On ranking. These are ranked by how long Meta has kept each ad running, not by impressions. Impressions are not obtainable: the Ad Library publishes no impression or reach figure for UK commercial ads (that is an EU political-ads feature, and it appears on 4 of our 1,255 cards), and the library's own sort by total impressions parameter is silently ignored, so grid position carries no impression signal either. We tested that directly: ascending, descending and no sort at all return identical card order. Days live is the only performance proxy available from outside an account, and an ad Meta has served for 150 days is one the advertiser keeps paying for.

The longest-running ad in the whole report, and every ad in the top six is a different argument. Five of the six are rechargeable-lighting product films. This is what a healthy top set looks like.

Five distinct arguments in the top six, three of them tied at 167 days. Un-produced phone-look creative is doing the heavy lifting. The one repeat argument is Throw it on, tuck it in, running at both 167 and 134 days.

Authority and material proof at the top, no offers anywhere. The one repeat is the Handcrafted in Devon provenance film, running as two ads at 119 days.

Six different creatives all making one argument: the same starter-kit discount, Was £73 Now £32.99, for up to 134 days. New assets, no new idea.

The most structurally complete top set: the garden-shed founder film (twice) alongside three catalogue product ads and a range film, all at 100 days. Story plus retargeting spine, which is the structure to mirror.

The top six are only three distinct creatives, one of them running four times, and the best is still just 76 days. Their 188-ad account has no durable winner and its top set is mostly one film repeated.

One 71-day ad and then a cliff to 9 days, all end-of-season sale creative. No evergreen spine at all.

Every top ad sits at exactly 36 days, so the account was rebuilt about five weeks ago. No evergreen exists here yet.

Top ad is 2 days old. This is a launch teaser burst, not a top set. Read nothing into this ranking.

All 83 creatives, grouped by content type and ordered by days live. Films show as three frames: opening, middle, close. Every image links to the live ad in the Meta Ad Library.

The single most durable type observed. Pooky and Naturalmat both use it. Hands, a mechanism, no narrator.

Styled room shot. Durable and cheap, but only when the room is genuinely aspirational.

Un-produced cropped shots. WNU evergreens 167 days on these. Cheapest durable asset in the set.

Only Piglet runs one. Builds meaning a discount cannot.

The DPA retargeting spine. Not a prospecting asset, but it should always exist.

One ad sells the breadth of the range instead of one SKU.

Only Naturalmat. Borrowed institutional credibility, and the most under-used lane here.

Real receipts (Strava, DM screenshots) instead of claims. Only Puresport does it.

Puresport and If Only If only. Durable for them because the offer is permanent, which is its own problem.

Biggest lane in the set. Works when the overlay states a hard USP, dies when it states a mood.

Material close-up. Good support, never the whole argument.

Widely run and mid-durability. Beautiful, but carries no claim, so it cannot do the convincing alone.

Puresport and Wild Nutrition only. Note it does NOT evergreen well here: 43-day median.

A statistic as the creative. One example in the set. Cheap to make.

Teach rather than sell. Wild Nutrition only.

Many creator clips in one frame. Social proof at volume in a single ad slot.

Demand as the argument. Needs a real sell-out to be honest.

Retail/community moment. TOAST only.

Caption-led in-joke. Cheap, scroll-stopping, and short-lived.

Piglet only. Hand-made stop motion. Distinctive and very cheap.

Lowest durability of any type here at a 7-day median. Produced fashion imagery churns fastest.

How it was pulled. Meta's official Ad Library API returns only political and social-issue ads for the UK, so it is unusable for commercial research. Everything here comes from rendering the public Ad Library in a headless browser, page-scoped by page ID, then splitting each account by media type. Creative assets were downloaded at full resolution from the ad archive rather than screenshotted, because the card preview is too small to judge styling.

Coverage. 1,330 unique ads against roughly 1,371 reported by Meta, about 97 percent. Shortfalls are concentrated in Wild Nutrition and TOAST, whose accounts are large enough to hit Meta's pagination limits. Those gaps affect ad counts, not the durability findings, because the long-running ads surfaced in every pass.

Rate limiting. The first pass was blocked partway through and returned false zeros for Wild Nutrition and Sézane: Meta serves a clean, healthy-looking page with the filters echoed back and the results silently dropped. Those were caught and re-run three and a half hours later once a control query came back green. Every number here was collected in a window that passed a control check at both ends.

What is judgement. 66 of the 83 content types were assigned after looking at that ad's own creative. The remaining 17 were inferred from copy, format and the brand's established pattern, and are marked "type inferred" in the library. Longevity comes from the harvest pass, not the ad detail page, because the detail view resets flight dates.

Snapshot. Live ads as at 23 August 2026. This is a point-in-time read of active creative and does not include ads that have already been switched off.