How to See Competitor Ads: Every Public Ad Library, Ranked

A public ad library is a platform publishing a list of the advertising it is carrying. At minimum that means the creative itself, the advertiser it belongs to and the date it started running. The better ones add the market it was shown in, the format, the destination address, the legal entity paying, and for regulated categories a spend range and an impression range. All of the major ones are free, most need no account, and between them they cover the channels where the great majority of a competitor’s budget goes.

What they do not contain is performance. There is no impression count, no click count, no conversion and no spend for an ordinary commercial ad anywhere in any of them, and there never has been. Every tool quoting those numbers for a normal advertiser is running a model over a traffic estimate, which is a different activity from reading a record. The libraries also stop at the platform’s own inventory: an ad bought directly from a publisher, inside a podcast, on connected television or through a retail media network appears in none of them.

1. 1Search the legal entity, not the brand. Libraries index the advertiser account, and the account is usually registered to a legal entity whose name differs from the brand on the website. Read the entity name off their terms of service or privacy policy first, then search that. Searching the trading name alone is the single most common reason somebody concludes a competitor is not advertising.
2. 2Set the country before you read a single ad. Every library filters by where the ad was shown, and the default is rarely the market you care about. Choose the country deliberately and record which one you chose. A competitor running twelve creatives in the United States and two in Germany is telling you where the budget sits, and you will miss it entirely if you read one market.
3. 3Sort by how long each ad has been running. Start dates are published and they are the closest thing to a performance signal these libraries contain. An ad live for four months has survived every optimisation decision the advertiser made in that time. A wall of creatives launched last week is a test. Rank by longevity and read the survivors first.
4. 4Collapse the set into variants rather than counting ads. Forty ads is usually four ideas tested ten ways, and the count is meaningless while the ideas are not. Group by headline and offer, then note how many variants each idea carries, because the idea they are testing hardest is the one they believe in most and are least sure how to express.
5. 5Open the landing page for every distinct idea. The creative is half the message and the destination is the other half. Follow each distinct ad through to where it lands, and note the offer, the form fields, whether pricing is shown and which page they chose. A competitor sending paid traffic to a comparison page rather than a home page is running a displacement play.
6. 6Check the categories the library treats differently. Political, electoral and social-issue advertising carries spend ranges, impression ranges and funding entities that commercial advertising never gets, and it is retained for years rather than weeks. If a competitor advertises on any regulated topic, that is where the numbers are, and it is worth checking even when you expect nothing.
7. 7Save what you found, because nothing here is an archive. A commercial creative disappears when the campaign ends and no rule obliges anyone to keep it. Screenshot the ads that matter, record the advertiser, the start date and the market beside each, and file them the day you find them. Everything else on this page depends on you having done that last quarter.

These libraries were not built for you. They exist because the law required them, and reading them well starts with understanding what the law asked for. European rules oblige the largest online platforms to keep a public repository of the advertising they carry, covering the content of the ad, the advertiser, whoever paid for it, the period it ran, the targeting parameters used and the reach it achieved, retained until a year after the ad last appeared, and reachable through both a searchable interface and a programmatic one.

That specification is why the fields are what they are. It is also why coverage is uneven: a platform below the size threshold has no obligation at all, and a platform above it has an obligation that attaches to European exposure rather than to advertising generally. A competitor advertising only in North America may be almost invisible in a library that a European competitor fills.

That list is the most useful thing anybody has written about judging an ad library, because it came from an enforcement action rather than a blog. Apply it as a five-question test to whichever library you are about to rely on. Can you search it by what the ad says rather than only by who ran it. Does it return results in the page rather than as a file. Does it show the whole creative. Does it name the entity that paid. And does it answer in seconds rather than minutes. A library failing three of the five is one where an absence of results tells you nothing.

The most expensive misunderstanding about this source is treating it as an archive. For an ordinary commercial ad it is not one. Retention is decided by the ad’s category and by where it was shown, not by the platform, which produces the counter-intuitive result that a single library is a live feed and a multi-year archive at the same time depending on what the ad is about.

This is a different problem from the one social feeds have, and it is worth keeping the two apart. A post leaves a feed because feeds are chronological and nobody scrolls back; the capture problem there is described under competitor social media. An ad leaves a library because a rule says it may, and the rule can change. The first is a property of how people read. The second is a policy decision by somebody with an interest in the outcome, and it moved twice in eighteen months.

The practical response is unglamorous and it is the whole of the discipline here. When you find a creative that matters, save it that day with the advertiser, the market and the start date beside it. Where a competitor’s landing page matters more than the ad, put a dated public copy of it somewhere a sceptic can open, which is what web archives are for.

The first thing most people do with a library is count. Forty active ads sounds like a competitor spending heavily and twelve sounds like one pulling back, and neither reading survives contact with how advertising is actually bought. Forty ads is usually four ideas tested ten ways, and the useful number is four.

Collapse the set before you analyse it. Group the creatives by the claim they lead with and by the offer they make, and you will typically find a small number of distinct propositions, each carrying a cluster of variants that differ only in image, headline order or call to action. Two things fall out immediately. The number of distinct ideas tells you how many audiences they believe they are selling to. The number of variants behind each idea tells you which one they are least confident they can express, because that is where the testing effort goes.

Rank by longevity, not by recency

Start dates are published and they are the only performance proxy these libraries genuinely support. Advertisers cut what does not work, so an ad running for four months has survived every optimisation decision made in that period and is a message the company has decided to stand behind. A dozen creatives launched in the same week is a test with no verdict yet. Reading the survivors first, and reading them as their considered position rather than as experiments, is the single change that most improves what you get from this source.

Read the destination as half the message

Follow each distinct idea to where it lands. A competitor sending paid traffic to a comparison page rather than to their home page is running displacement and expects the visitor to already be evaluating somebody. A destination with pricing on it means they are comfortable competing on price; one with a form and no pricing means the opposite. The parameters in the destination address often name the campaign, the audience and occasionally the segment, in the competitor’s own internal vocabulary.

Watch the entry claim change

Because start dates are attached, the set is a dated record of what a competitor thought was persuasive at a point in time. Comparing your capture from last quarter against today shows a repositioning long before the website catches up, since paid copy is cheap to change and a home page relaunch is not. A category word appearing in ads before it appears anywhere else is the earliest reliable sign that a rival is moving.

1. Meta Ad Library

The most generous of the major libraries for commercial advertising, and the only large one that lets you search by keyword as well as by advertiser, which means you can find rivals you did not know were bidding on your category. Covers their whole app family. Set the country and the category first, because both default to something unhelpful.

2. Google Ads Transparency Center

One advertiser view spanning search, video and display placements, with formats and regions. Its distinguishing feature is identity: advertisers complete a verification process and the resulting name, location and paying entity are displayed. That turns the library into an entity lookup as well as a creative archive, which is worth more than most people realise.

3. LinkedIn Ad Library

Sponsored business content by advertiser, going back to June 2023. This is the channel none of the consumer libraries reaches and where a large share of business software advertising actually runs, so for a B2B competitor it frequently holds more of the picture than the two larger libraries combined.

4. TikTok Commercial Content Library

Advertising shown in Europe, with the creative, the advertiser, the run dates and the declared targeting parameters. The targeting field is unusual and useful: it is the advertiser stating who they were trying to reach, rather than you inferring it from the creative.

5. The X ads repository

It exists, it covers European inventory, and a regulator found it inadequate in December 2025. A rebuild is under way under supervision, committing to content and targeting search filters, in-page results, faster responses, full creative and destination addresses, and programmatic access. Until that lands, treat an empty result here as uninformative rather than as evidence.

6. Snapchat’s political and advocacy ads library

Published as downloadable files rather than as a searchable interface, covering political and advocacy advertising with spend and impressions attached. Narrow in scope and awkward in form, and worth knowing about precisely because the numbers exist in a category where numbers usually do not.

7. Marketplace and retail media repositories

The same European obligation applies to the largest marketplaces, which is the only reason retail media advertising is visible anywhere. If your competitors buy sponsored placement inside a marketplace, this is the channel that would otherwise be entirely dark to you. Reach each one from the platform’s own transparency pages, since the interfaces are not consistent and several are new.

8. Election and political ad archives

The one category with real numbers: spend ranges, impression ranges and the funding entity. Worth checking even for a competitor you assume has nothing political to say, because social-issue advertising is defined broadly and companies advertising on regulation, employment or sustainability topics frequently land inside it without realising.

9. The landing pages behind the ads

Half the message, and the half that is not in any library. The offer, the form fields, whether pricing is shown, and the campaign parameters in the address. Open one for every distinct idea rather than for every ad, and record the page they chose alongside the creative.

10. Your own dated captures

The only copy of a commercial creative that survives the campaign ending, and therefore the one thing here that survives on your terms rather than theirs. Four fields beside each capture is enough: advertiser, market, start date, offer. A year of that is a messaging history nobody else in your category has, assembled from a source that does not keep one.

11. Advertising inside AI assistants

The newest surface and the one with no library at all. Sponsored placements began appearing in a major assistant during 2026, initially in the United States for users on free tiers, clearly labelled and separated from the response, with no public repository of advertisers or creatives. Everything anybody knows about who is advertising there comes from people observing it themselves, which is exactly the position every channel was in before the transparency rules arrived.

• All the major ones are free. No subscription, and in most cases no account either. This is the least gated material in competitive marketing research, which sits oddly beside the number of products selling access to reformatted versions of it.
• The country filter is the setting that decides your results. Every library scopes by where the ad was served. Choose the market deliberately, record which one you chose, and check at least two if the competitor operates internationally, because the difference between them is a budget allocation you can read directly.
• Search behaviour differs more than the interfaces suggest. One major library supports keyword search across creatives, which finds advertisers you did not know to look for. The others are advertiser-first, so you must already know the name. Use the keyword-capable one to discover and the others to confirm.
• Programmatic access exists and is not the point here. Some libraries publish interfaces for researchers with their own approval processes and terms. For competitive work on a handful of rivals, a browser and twenty minutes a month produces everything you need, and the reading is the part that creates the value.
• Do not click a competitor’s live ad to inspect it. It costs them money, which is petty, and it puts you into their retargeting audience, which distorts what you see afterwards. The library shows the creative without either consequence, which is one of the reasons to work there rather than in the feed.
• An empty result is ambiguous, not negative. It can mean they are not advertising, that they advertise under an entity you have not found, that they advertise in a market you did not select, or that the library is one of the ones that does not work well. Rule out the first three before reporting the fourth.

Reading a public ad library raises nothing at all. Every ad in it was published to strangers, at the advertiser’s expense, in order to be seen. The exposure on this page is unusual because it sits almost entirely on the other side: what you do in your own advertising once you have read theirs. None of this is legal advice.

• Naming a competitor in an ad is generally allowed, under conditions. European law permits comparative advertising where the comparison is objective, covers products meeting the same needs, compares verifiable and representative features, does not mislead, does not discredit the rival and does not create confusion between the two brands. The exposure is in the copy, not in the decision to compare.
• Every claim about a rival needs a dated source you could produce. The practical test is whether you could put the evidence on a table tomorrow. Superlatives with no measurement behind them, and comparisons against a version of their product that no longer exists, are the two ways these claims are lost.
• Do not copy their creative. An ad is a copyrighted work and their brand assets are trademarks. Studying a competitor ’s structure, offer and claim is research; reproducing their layout, imagery or wording is an infringement claim with your own campaign as the evidence.
• Impersonation is a different order of problem. Running ads that a reasonably observant person could take for the competitor’s own, whether through their name in the headline, their brand colours or a lookalike destination, moves from comparison into confusion, which is the thing the rules are actually built to prevent.
• Do not click their live ads as a research method. Repeatedly clicking a competitor’s advertising to see where it goes spends their budget and, done systematically, is the sort of conduct platform terms treat as invalid activity. The library exists so that you never need to.

• What any of it cost, for a commercial ad. No spend field exists outside the political categories. Proxy: their filed sales and marketing expense as an audited ceiling, and your own auction reports for how hard they are bidding against you specifically.
• Whether any of it worked. No impressions, no clicks, no conversions. Proxy: how long each ad has run, which is a revealed preference by the only party with the data.
• Anything bought outside a large platform. Direct publisher deals, podcasts, connected television, newsletters, events and most retail media carry no repository. Proxy: the sponsorship and partner logos on the publications and events your buyers actually consume, checked twice a year.
• A campaign that has already ended. Commercial creatives are not kept, and one archive has been removed outright. Proxy: your own captures from last quarter, which is why the saving habit is the whole discipline here rather than a nice-to-have.
• Which segments actually saw it. Where a targeting field exists it records the advertiser’s declaration rather than delivery. Proxy: read the creative itself, since an ad written for a job title names that job title, and build the pattern in a messaging framework rather than in a slide.

Monthly is the right cadence for the set, because campaigns generally run in weeks rather than days and a fortnightly check mostly returns the same creatives with a different sort order. Work one competitor at a time, capture the distinct ideas rather than every variant, and stop when you have four fields for each: market, start date, offer and landing page.

Four events justify going in immediately rather than waiting for the month to turn. A funding announcement, because new money reaches paid channels before it reaches anything else. A product launch, because the launch window is the one period where a competitor cycles through a dozen variants in a fortnight and the sequence of what they tried and dropped says more than the survivor does. A pricing change, because the entry price usually appears in an ad before it appears on a pricing page. And any quarter where your win rate against them moves, since the explanation is often sitting in a creative nobody read.

Keep the record outside the library. That is the discipline this source demands and the one most teams skip until the first time somebody asks what a rival was claiming eight months ago and the honest answer is that nobody knows. Your own dated captures are the only version of this history that is guaranteed to exist next year, and building them costs twenty minutes a month.

The thing that goes stale on this page is not the information, it is the habit of collecting it. Eleven interfaces, each with its own filters and its own idea of what an advertiser is called, most of them returning nothing new in any given month: that is the exact profile of a task somebody does diligently in January, twice in the spring, and not at all after the summer. Nobody decides to stop. It just becomes the thing that gets skipped when a launch lands, and the skipping is invisible until somebody asks what a rival was claiming two quarters ago and there is no answer.

Keeping a continuous, dated record of what a competitor is saying, so that the history belongs to you rather than to whoever is obliged to publish it this year, is a standing job, and competitive intelligence software is the category that takes it on. Flares tracks messaging and campaign changes as they appear and stamps each one with the date it was first seen, which is the part a library will not do for you. What it cannot do is tell you what the change means. A rival replacing the word platform with the word agent across their entire creative set is a fact a system can hand you in a morning; deciding whether that is a category bet worth answering, or a phase worth ignoring, is a judgement about your market that stays with the person who knows it.