The examples
20 creatives from 98 advertisers, as the public libraries returned them. Click any to open it here.
Benchmarks
| Company | Channels | Ads observed | Meta | Longest run | ||
|---|---|---|---|---|---|---|
 Canva |    | 29,812 | 565 | 560 | 759 | 212d |
 ElevenLabs | | 14,573 | 462 | 831 | 918 | 320d |
 Rippling | | 7,628 | 565 | 554 | 631 | 182d |
What the data shows
Message Ads are the quiet story in this data
18 advertisers in this set run Message Ads sent from a named human rather than the company page, including Cursor, Replit, Grammarly, Perplexity, ElevenLabs, Runway, Clay, Warp, Descript, Photoroom, Bland, Vapi, Decagon, Speak, Luma, Lindy, Canva, Rippling. This is a different acquisition motion from everything else on the page: the ad lands in an inbox, reads as outreach, and carries a personal sender name and photo rather than a logo. It is also invisible to anyone judging a competitor by their feed. Several of these advertisers run large, highly visible Meta libraries and almost nothing in the LinkedIn feed, while quietly running hundreds of InMail sends. A company that looks absent from LinkedIn on a casual check may be its heaviest user.
The ad-type mix is wider than any other channel
Across this set the observed types run Single Image Ad (14338), Video Ad (6598), Document Ad (1166), Message Ad (1032), Carousel Ad (361). That variety is structural: LinkedIn supports document ads, article ads and conversation formats that have no equivalent on Meta, and B2B advertisers use them for genuinely different jobs. Document ads carry a gated asset and are a lead-capture format. Article ads borrow the credibility of editorial. Single image remains the volume workhorse. Reading a LinkedIn library by ad type tells you which stage of the funnel an advertiser is actually spending on, in a way a format breakdown on Meta cannot.
What LinkedIn discloses, and what it does not
LinkedIn's targeting disclosure is the single most valuable field in any public ad library, and it is also inconsistent. In this scan every ad carrying targeting published location and language, and none published job title, seniority or industry, the fields that would actually reveal an ICP. Treat any tool claiming to show you a competitor's LinkedIn job-title targeting with suspicion: the field exists in the schema, but it is not reliably populated. Impression bands are more dependable, and are published as ranges such as under 1k or 1k–5k rather than exact counts.
B2B creative reads differently because the buyer is different
The copy in this set is markedly longer than on Meta, and the reason is the read context: a LinkedIn feed browser is at work, in a professional frame of mind, and will tolerate a paragraph. Proof appears earlier: named customers, funding, headcount, because the reader is evaluating whether the vendor is safe as much as whether the product is good. And the calls to action skew toward demo booking and asset download rather than signup, which reflects a sales-assisted motion sitting behind the ad. An advertiser running self-serve signup CTAs on LinkedIn is usually a product-led company testing whether the channel works at all.
Do this yourself
Sprites builds ABM audiences from your ICP, writes the thought-leadership creative and runs the retargeting that follows the buying committee.
manage LinkedIn Ads with AI →Every advertiser featured here
Run ads like these, without the team
Sprites reads competitor libraries the same way this page was built, then builds and runs your campaigns against what it finds.