Dataset · updated 2 October 2026
What 41 D2C brands' public ad libraries actually contain
Across 41 brands we decoded from the public Meta Ad Library, live library size ranges from 3 ads to 2,247. In the 35 reads where every advertiser is matched to its own figures, we have decoded 13,842 distinct creatives in total. The gap between a brand and its closest rival is routinely 10× or more — one lending brand was running 3 live ads while a direct competitor ran 940. And library size is not the same as creative variety: in one case a 300-ad library carried just 37 distinct creatives.
Everything here comes from the public Meta Ad Library, public Instagram and public comments. None of it comes from any brand's ad account, so there are no spend, conversion or cost-per-order figures on this page — those are not observable from outside and we do not infer them. Brands are described by category rather than named.
The corpus in totals
These are the figures for the whole set rather than any one brand. Each one is a sum over advertisers that have been individually matched to their own counts, which is a recent and non-trivial distinction — for most of this corpus's life we had the numbers without knowing reliably whose they were.
| Measure | Total | What it counts |
|---|---|---|
| Distinct creatives decoded | 13,842 | Across 35 attributed reads and 99 advertisers, after deduplication within each pull |
| Live ads observed | 17,196 | Only where a live total was measurable — mixes brands and rivals, so not a corpus size |
| Live ads with an exact size | 4,528 | Across 26 libraries whose size never exceeded our pull limit, so the count is real rather than a ceiling |
| Reads with no live total | 16 | Decoded counts only; the Ad Library returned no active-ad figure for any advertiser in them |
| Pulls that hit a ceiling | 8 | One at our 1,000-ad brand cap, seven at a 150-creative rival cap |
| Brands with no readable data | 6 | Excluded entirely rather than recorded as running no ads |
Source: our own 35 attributed competitor reads, re-audited 2 October 2026. The creative and live-ad totals were each reproduced independently from the per-advertiser table before being published here.
The 17,196 needs its caveat stated rather than buried: it counts every advertiser we measured a live total for, brands and rivals together, and it excludes 16 reads that carried no live totals at all. It is the volume we have observed, not the size of any market. The 4,528 is the stricter figure — live ads in libraries whose size is exact rather than capped, which is the only subset where a deduplication rate can be computed — and then only over the ads actually fetched.
How large a live library gets
| Category | Live ads | What stood out |
|---|---|---|
| Apparel (India) | 2,247 | The largest library in the set, and it belongs to a rival rather than a brand we work with. Sampled at 402 creatives decoded, not read whole. |
| Vitamins (UK) | 1,034 | Also a rival's library; sampled at 150 creatives decoded. |
| Beauty marketplace (India) | 1,000 ads | 1,258 distinct creatives decoded — the deepest read in the set. |
| Grooming (India) | 1,047 | 475 creatives decoded. |
| Women's nutrition (UK) | 536 | Ahead of both rivals, at 338 and 107. |
| QSR franchise (US) | 300 | Only 37 of them distinct. See below. |
| Supplements (India) | 279 | Against rivals at 874 and 972. |
| Intimates (US) | 104 | 76.8M Instagram followers, and 2.7× less paid volume than smaller rivals. |
| Vitamins (UK) | 61 | Against rivals at 255 and 1,034. |
| Feminine care (India) | 24 | Against rivals at 471 and 210. |
| Baby care (India) | 18 | Against a category rival running 314. |
| Wellness resort (India) | 12 | In a category that barely advertises: rivals at 145 and 7. |
| Lending / fintech (India) | 3 | Its entire live library, against a direct rival running 940. |
Source: Galvor's own decode of 41 brands' public Meta Ad Library entries, read between 2026 and October 2026. Live counts are as at the date each library was read; an ad library moves daily.
Library size is not creative variety
A US quick-service franchise was running 300 live ads carrying only 37 distinct creatives. Of the 352 media files behind them, 315 were byte-identical repeats — 89.5%: 146 of 156 videos and 169 of 196 images. Almost certainly the same assets re-uploaded across franchise locations.
This matters for anyone reading a competitor's library as a signal of creative investment. A raw count of live ads can overstate how much distinct creative a brand is actually testing by more than an order of magnitude, and a competitive read that does not de-duplicate will reach the wrong conclusion.
Audience size does not predict paid volume
A US intimates brand with 76.8 million Instagram followers was running 104 live ads. Two smaller rivals were running 288 and 267 — about 2.7× the paid creative volume on a fraction of the audience. We read 67,712 comments across the set.
Organic reach and paid creative volume are independent decisions, and the brands with the most followers are not necessarily the ones testing the most creative.
What was read, and how
- 41 brands, each decoded on the same vocabulary — hook form, narrative framework, CTA type, shot timing and format mix. See the attribute taxonomy.
- Two named competitors per brand, decoded on the same vocabulary, so the comparison is like for like rather than a brand's library against an impression.
- Organic Instagram and comment analysis alongside the paid read — in one case 178 organic posts over a year, in another 67,712 comments.
- Live totals and decoded totals are different numbers, and both are recorded. The lending rival above had 940 live ads; 40 of them were pulled and decoded. A library size is a count the Ad Library reports; a decode describes only the ads we actually fetched, which is never more and is sometimes fewer.
- Exact or capped is recorded per library. A library that never reached our pull limit has an exact size; a larger one is capped and says so. An exact size does not mean every ad was downloaded — ads expire between the count and the pull — so the number read is recorded separately.
What is not here: spend, impressions, conversions, cost per order, CAC or return on ad spend. None of those are visible in a public ad library. The longest an ad has been running is visible, but we do not treat a long-running ad as evidence that it converted — a brand continuing to pay for a creative is not proof that it worked.