LinkedIn Ad Library: How to Use It (and What It Can't Show You)

The LinkedIn Ad Library is LinkedIn's public ad transparency tool. It lists the ads each advertiser is currently running on the platform, along with who paid for them and, where LinkedIn discloses it, the targeting parameters and rough impression ranges behind them. It is free, it needs no account, and it covers every advertiser on LinkedIn, not a curated selection.

It exists because ad transparency regulation requires it, and that origin explains its shape better than any feature list does. It was built to answer a regulator's question, "who is paying to say this, right now," not a marketer's question, "what works in my category." Everything it does well and everything it refuses to do follows from that one distinction.

For the job it was built for, it is the best source there is, and no third-party tool can beat it on its own ground. The ads come straight out of LinkedIn's ad serving system, so there is no question of authenticity, no sampling and no lag. An ad is listed because it is serving impressions today.

That immediacy is genuinely valuable, and it is worth saying plainly: no dataset assembled by anyone else, AdPeekly included, can tell you about a campaign that launched an hour ago. If your question is "what is this one company running this week," stop reading and go use the official tool.

linkedin.com/ad-library

Ask it a research question instead of a compliance question and it runs out of road quickly. The limits below are not bugs or missing features on a roadmap; they are what a transparency register looks like when you try to use it as a competitive research database.

When a campaign ends, its ads leave the library. Not archived, not flagged as inactive: gone, with no way to look them up again. And while an ad is live, the library never tells you when it started, so you cannot distinguish a creative launched yesterday from one that has been running since January.

Those two facts together remove the single most useful signal in competitive research. Duration is the closest thing to a public performance metric that exists in advertising: nobody keeps paying, month after month, to serve a creative that does nothing. A tool that deletes ended campaigns and hides start dates is a tool that can show you what is running, but never what worked.

There is no industry filter, no format filter and no way to ask a question broader than a single advertiser. "Show me what cybersecurity companies are doing" is not a query the library accepts, so scoping a category means listing the companies yourself and searching them one by one.

What you get back is noisier than it looks, too. Advertisers run one creative in many near-identical variants, different headline, cropped image, swapped call to action, and the library lists each variant as its own ad. A single campaign can read as a dozen, so counting ads in the library tells you how much an advertiser is testing, not how many ideas they actually have.

There is no export, no CSV, no API. Every ad you want to keep for a swipe file or a competitive deck gets screenshotted by hand, and anything programmatic, a dashboard, an alert, a question asked inside an AI assistant, is simply off the table. The data is public to look at and private to use.

That history exists, and keeping it is what AdPeekly does. AdPeekly is a research library of B2B LinkedIn advertising: it revisits the same public Ad Library on a repeating schedule, records every ad it finds across a curated list of 649 companies, and keeps each one after LinkedIn drops it, stamped with the first and last date it was seen running. 253,610 published ads later, the questions the section above could not answer have answers. Here are three of them, and what each one is good for.

Most LinkedIn ads are short. A third of them are gone inside a week, and the median ad runs 19 days. The interesting part is the tail: roughly one ad in fifteen is still running after three months, and those are the creatives worth studying, because they are the ones their owners kept paying for. Finding them is a sort, not a search: order the library by observed duration and the survivors come to the top, which is the single most useful thing you can do with a dataset that remembers.

Industry moves the number far less than you would expect, and one vertical is a real outlier. Ten of these twelve industries sit between 14 and 21 days, clustered within a few days of the 19-day dataset median. Then marketing and growth companies keep their median ad live for 31 days, roughly 60% longer than everyone else, and operations and supply chain cycle theirs out fastest at 14. The people selling marketing tools run the longest campaigns in B2B, which is either confidence or a longer consideration cycle, and this dataset cannot tell you which.

The practical reading is the opposite of what a spread this narrow suggests at first glance. If your campaigns run three weeks, you are doing what almost every B2B advertiser does, whatever your vertical. The industry filter is still worth having, but for finding the outlier and for seeing which companies and formats sit behind a number, not because each vertical follows a different rule.

Two thirds of what looks like volume is repetition. Grouping visually matching creatives and keeping one representative of each, those 253,610 ads collapse to 95,893 distinct creatives. Search results and every count on this site use the deduplicated figure, because a gallery of 200 ads that turns out to be 40 ideas in five variants each is the failure mode readers rightly punish. The rule is written out at /methodology. For anyone building a swipe file, that rule is the whole point: fifty saved ads should be fifty ideas, not ten ideas saved five times.

One caveat worth stating before you compare these numbers to your own count on LinkedIn: this dataset is a curated list of 649 B2B companies, and only ads that passed review are published. It is deliberately not a live unfiltered mirror of everything LinkedIn holds, so the count for any single company will be smaller than the raw total you could tally by hand. Every published ad has been checked, not just collected.

If what you actually want is to see the ads rather than read about the tool that lists them, the LinkedIn ads examples page is the shorter route: forty real creatives across every format, and the measured format mix behind them.

And if the question behind all this is specifically about a competitor rather than the tool itself, the guide to finding competitor ads on LinkedIn walks through the exact search steps, what the Sponsored badge does and doesn't confirm, and how to catch ads that never make it into this library at all.

It is the tool those three findings came out of, built for the people who were doing this research by hand: B2B marketers, founders and agencies who need to know what a category is actually advertising before they write their own brief. Where LinkedIn's Ad Library gives you one advertiser at a time with no past tense, AdPeekly gives you 95,893 distinct creatives from 649 curated B2B companies, every one of them carrying the dates it ran, the format, the industry, the funding stage and the company size.

In practice that means the three moves above are one interface away. Filter to your category and read its real benchmark. Sort by duration and pull out the ads that survived a quarter. Save the deduplicated winners into a board and export the swipe file you would otherwise have screenshotted by hand. And because the same dataset is exposed to AI assistants through AdPeekly's MCP server, you can also just ask the question in Claude or ChatGPT and get the ads back in the conversation.