Taboola Ad Library: What Exists and How to Research Taboola Ads
Taboola publishes no official ad library. How an independent archive of captured Taboola ads works, what it records (advertisers, observed spans, geo, landing pages), what it can't show, and which features are free.

Search for "Taboola ad library" and you won't find an official one. Meta has an official ad library. Google has a Transparency Center. TikTok has a Commercial Content Library. Taboola, one of the largest native advertising networks, offers no public, searchable library of the ads running through it: its Trust Center publishes Digital Services Act transparency reports and disclosure rules, but no ad archive (checked 7 October 2026). Several third-party ad intelligence tools, including OpenAdLibrary, which publishes this blog, archive Taboola ads they capture; we have not tested other tools' Taboola coverage for this article.
Our own archive is one of them. In a July 2026 snapshot it held 171,050 Taboola creatives, 167,524 of them (98%) first captured in the preceding 30 days; that is a historical measure of our collection, not the size of Taboola's market or a current count. This guide covers why no official Taboola library exists, exactly what our Taboola index captures, how to use it for competitive research, and where its honest limits are. If you want the workflow-focused walkthrough of researching competitors on the network, that lives in our Taboola ad spy guide and the /spy/taboola tool page; this article is about the library itself: what is in it and how to read it.
Why Taboola has no official ad library#
The big transparency surfaces you already know exist because regulators forced them or because political-ads scrutiny made them unavoidable. Meta's library grew out of election-interference fallout. Google's Transparency Center and TikTok's Commercial Content Library expanded under the EU's Digital Services Act, which obliges very large online platforms to run public ad repositories.
Taboola is not on the European Commission's list of designated very large online platforms and search engines (last updated 31 August 2026), so the DSA's repository duty does not apply to it. Taboola does require every sponsored item to name the sponsoring company, but it offers no archive you can search. The result is a genuine gap: you can look up a brand's active ads in Meta's Ad Library, but the network serving ads across thousands of open-web publishers has no equivalent official lookup. We wrote about this gap more broadly in What Is a Native Ad Library and Why One Didn't Exist Until Now.
That gap is what an independent library fills: instead of the network publishing its own ads, a third party observes the public placements and archives them. The ads are shown publicly, so an archive records what any visitor could see, much as clipping services archived print ads. That describes the method; it is not a legal clearance for every reuse of a captured creative. Whether copying or republishing one is lawful depends on copyright and trademark law and the facts, and US courts decide fair use case by case.
What the OpenAdLibrary Taboola index captures#
We capture public Taboola placements, store the creative at full quality, and trace what sits behind each ad. Where the data is available, an entry includes:
The creative itself: the full-quality image plus headline, not a degraded thumbnail. The advertiser: the brand label behind the ad, resolved from the tracking chain where possible rather than taken from the widget byline (real-advertiser resolution for native ads is a Founders feature). Observed dates: first-seen and last-seen dates from our captures. They show when we observed an ad, not whether it was profitable; our ad longevity guide discusses the signal. The landing page: we follow the redirect chain to the destination page or pre-lander without clicking live ads. Our July 2026 snapshot recorded 1,081,997 traced landing pages across all networks, a historical collection figure. Geo and device: the countries and device types (desktop, Android, iOS) where we captured each creative, which reflects where we looked as well as where the advertiser bought. Vertical classification: creatives get an industry category when the classifier is confident, so not every creative has one. With the Founders plan you can filter by category, geo and device, for example insurance ads captured in Australia.
Here is a historical entry from July 2026: a health creative we had observed over 19 days. That observation window says nothing about its economics or whether its claim is accurate:

Not every captured Taboola advertiser is a direct-response or arbitrage buyer. Large brands appear too, for example this HSBC credit-card creative, which we had observed over 21 days at capture:

What our July 2026 Taboola snapshot contained#
The library is queryable, so we can summarize our own collection rather than cherry-pick examples. It is a sample of what we captured, not a census of Taboola. The vertical mix of classified Taboola creatives in our July 2026 snapshot:
| Vertical | Classified Taboola creatives in our July 2026 snapshot (collection coverage, not market size) |
|---|---|
| Health | 6,857 |
| Finance | 5,936 |
| Insurance | 4,961 |
| Ecommerce | 3,688 |
| Home & garden | 2,961 |
| Software | 2,395 |
Health, finance and insurance had the most classified creatives in that snapshot. Creative counts show what we captured and classified; they do not measure spend, profitability or how hard a vertical is to compete in, because one advertiser can run thousands of variants and another a handful. If you want the advertiser-level view, Who Advertises on Taboola? breaks our collection down by advertiser and vertical.
Advertiser-level counts need the same care. In the July 2026 snapshot, the resolved advertiser labels with the most distinct Taboola creatives included Yahoo Search (4,184), OTTO Insurance (2,337) and Comparisons.org (1,095). A large number of creative variants does not show profit, spend or reach: it can reflect automated variant generation, many affiliates or partners running under one brand label, or how often we captured that advertiser.
Observed run lengths need the same care. The run-length tier percentages previously published here are withdrawn because their tier definitions and source extract are not available. A short observed run can mean a paused test, a rotation or a gap in our capture; a long one can mean steady delivery, a brand campaign, a slow test or an ad nobody switched off. Longer-observed creatives, like the hearing-aid ad above at 19 days, are useful for studying messaging and funnels, not as proven winners, and an archive is one practical way to find them.
How to use the Taboola ad library#
The library is a research surface, not a scrolling feed. The highest-leverage workflows:
Search an advertiser (Founders plan). Type a competitor or any brand you have seen in a feed and get every creative we have captured from them on Taboola: headlines, images, first-seen dates, landing pages. This shows the part of their activity our capture observed, which is broader than what you happened to see but not necessarily their whole playbook. Filter a vertical and geo (Founders plan). Pull, say, insurance ads captured in Australia and sort by days running. The top of that list shows the angles observed longest in our capture; treat them as hypotheses to test, not proven angles. Trace the funnel. Open a captured creative's redirect trail and landing page, which the free plan includes for native ads. The ad is the hook; the pre-lander and offer page are where the sale is attempted. Our guide to reverse-engineering a competitor's native ad funnel walks the full method. Watch for movement. New creatives from a known advertiser can signal a new test or offer. A creative that stops appearing after weeks may have been paused for performance, budget or seasonality, rejected after review, or simply missed by our capture. Pull it programmatically. Library data is available through our REST API and hosted MCP server on the Founders plan (see our native ad data API), so you can pipe Taboola creatives, advertisers and landing pages into your own tooling or AI agents.
If you are new to how the network itself functions — bidding, placements, the publisher side — read How Taboola Ads Work first; the library makes a lot more sense once you know what a "placement" actually is. And if you are deciding between Taboola and its main rival, our Taboola vs Outbrain data comparison uses this same index to compare the two networks head to head.
A worked example: researching one advertiser end to end#
Abstract workflows are easy to nod along to, so here is what the library produced on a single query in July 2026. Take OTTO Insurance, the auto-insurance comparison advertiser we mentioned above.
In July 2026, searching the name returned 2,337 distinct Taboola creatives attributed to the brand, with the longest-observed ones spanning the full 30 days of our observation window at the time. Scanning the headlines, the angle taxonomy is immediately visible: mileage-based hooks ("drivers who drive less than X miles"), age-bracket hooks, state-specific urgency hooks, and seasonal rate-change hooks. The number of variants shows how much creative this advertiser rotated in our capture; it does not show conversion rates, spend or which variants worked.
Then the landing-page traces. Following the captured redirect paths shows which headlines route to which pre-landers, where the quote-form funnel begins, and how the compliance language shifts between geos. In one sitting you can assemble the advertiser's observable creative and funnel structure without screenshotting feeds on different devices and VPN exits. It won't tell you which parts are profitable or compliant. Repeat for the top five advertisers in your vertical and you have a map of competitors' observed activity before you commit ad spend.
The same query pattern works defensively. Brands can search their own name to look for affiliates using unapproved claims or copycats using their name. An archive helps because native ads can stop running before anyone records them; a dated capture with its landing-page trace is a record of what we observed, and whether it is usable as evidence in a dispute is a question for the brand's lawyers.
What the July 2026 "98% first seen in 30 days" figure meant#
One number in this article deserves unpacking: in our July 2026 snapshot, 167,524 of 171,050 Taboola creatives (98%) had been first seen within the previous 30 days.
A figure like that has more than one explanation. Advertisers do rotate native creatives, and Taboola itself advises introducing new creative regularly, but a high share of recently first-seen creatives can also reflect changes in our own capture coverage and in how a creative is defined. It shows that the July 2026 collection was recent, not how fast advertisers kill losing variants. Recency matters because angles, advertisers and offer terms change, so check the dates on any entry you rely on.
The flip side is a trade-off we state openly: our collection is strongest on recent history, not multi-year archaeology. We cannot show you what ran on Taboola in 2023, and no third-party archive can be complete for past years. Recent captures show what was running recently; they cannot tell you what is winning.
Honest limitations#
An independent library observes the network from the outside, and you should know exactly what that means:
It is a deep sample, not a census. We capture ads by repeatedly observing public placements across publishers, geos and devices. The July 2026 snapshot of 171,050 creatives was a large window into Taboola, but no third party sees every impression on every publisher in every country. If an advertiser runs only on a handful of niche publishers in a geo we observe lightly, we may catch them late or not at all. Longevity is observed, not lifetime. "Running 21 days" means we observed the creative over 21 days, and in July 2026 our observation window extended to roughly a month. An ad may have started before our first capture or continued after our last. Treat longevity as a floor, not an exact age, and not as a performance measure. No spend or performance numbers. We show what was captured, who is behind it, how long we observed it and where the click goes. We do not show CTR, conversions or budgets, and we do not publish "estimated spend" figures; any tool showing precise Taboola spend per advertiser is modeling, not measuring. Geo coverage is weighted. English-language markets were our deepest in the July 2026 snapshot (251,633 English creatives across the index), with strong German, Spanish and French coverage. Thinner geos have thinner data.
We would rather you know these edges than discover them mid-research. The mechanics of how capture works — and why we never click live ads — are documented in How Ad Spy Tools Capture Native Ads.
Taboola ad library vs. a Taboola ad spy tool#
The two phrases overlap but the intent differs, and we maintain both surfaces. The library is the archive of what we have captured. The Taboola ad spy tool is the workflow on top of it: search and filters, saved research, funnel tracing and real-advertiser resolution for native ads, and Creative Studio, where you can remake native image ads with AI (all on the Founders plan). Use competitors' ads for research, not as material to copy. If your question is "what is running on Taboola," start with the library. If your question is "how do I compete with this specific advertiser on Taboola," start with the spy tool.
Either way, a free account (no card) shows the 20 newest Meta, TikTok and native ads, with ad detail, redirect trails and captured landing pages for native ads. Searching for an advertiser requires the Founders plan ($29.99/month).
The bottom line#
There is no official Taboola ad library. Taboola is not a designated very large online platform under the DSA, and its Trust Center publishes transparency reports rather than an ad archive. The practical substitute is an independent archive built from captured public placements. Ours held 171,050 Taboola creatives in a July 2026 snapshot, 98% first seen in the preceding 30 days, with the advertiser, observed dates, geo, device and traced landing page where available. That is enough to answer useful research questions about Taboola: who is advertising, what they say and where the click goes. It cannot tell you what is profitable.






