Taboola Statistics: Company Figures and What a June 2026 Ad Capture Showed
Taboola's reported 2025 company figures, and a dated June 2026 snapshot of 206,145 captured Taboola ads by vertical label, cross-network count and observed runtime, with the limits of each.

OpenAdLibrary's June 2026 collection snapshot contained 206,145 captured Taboola creatives, out of 725,882 creatives captured across 49 network labels. Those are counts of what our capture collected, not measures of Taboola's size, inventory share or advertiser spend, and the extract behind them cannot be reproduced today, so treat every collection figure on this page as a dated historical snapshot. This study sets out what that snapshot recorded (scale, vertical labels, cross-network counts and observed runtimes) alongside the company-level figures Taboola itself reports.
Where these numbers come from#
Taboola does not publish ad-level data, and we could not find an official public ad library of the kind Meta and Google run (checked 7 October 2026); that gap is why we built the independent Taboola ad library. Taboola does report company-level figures. Its annual report for 2025 shows revenue of $1.91 billion, about 2,200 Scaled Advertisers (advertisers with more than $100,000 of gross spend over the trailing four quarters) in the fourth quarter of 2025, about 14,000 digital property partners and a reach of over 600 million daily active users. What those filings cannot show is which ads run. Observation adds that: capturing public placements, storing each captured creative with its resolved advertiser where possible, recording the geo and device of each observation, and tracing the landing page through the redirect trail without clicking the live ad.
A few words on what "a creative" means here, because the definition drives every count below. Captured ads are deduplicated on the creative itself — the image and headline — so a single ad seen across fifty publisher sites counts once as a creative and fifty-odd times as observations. Advertisers are resolved by following each captured ad's redirect trail to its final landing domain, without clicking the live ad, rather than trusting the display name; that is how the collection separates a brand from the tracking domains in front of it. In this study, "live" meant observed within the June 2026 collection window. It was a sample of what our capture saw at the time, not a census of everything running on Taboola.
Creative volume is also worth calibrating as a metric. It is not spend: one creative can carry a large budget, and ten creatives can carry a small one. We have no measurement of how closely creative counts track spend, advertiser activity or competition, and counts also depend on where and how often our capture ran on each network. So the figures below describe our collection, not how much advertiser activity a network hosts or how contested a vertical is.
Three honesty notes before the numbers. First, an independent collection counts only what it observes, so every figure here describes our June 2026 collection coverage, not Taboola's totals or market size. Second, vertical labels cover only the classified subset of creatives, so vertical counts are partial. Third, the runtime figures are clipped by an observation window of roughly 38 days and measure the span between first and last capture, not paid delivery, so the true runtimes of the longest-observed ads may be longer. For the cross-network picture these numbers sit inside, see native advertising statistics 2026.
What the June 2026 collection captured (statistics 1–8)#
In the June 2026 collection snapshot: (1) we had captured 206,145 Taboola creatives; (2) that was 28% of the 725,882 creatives captured across 49 network labels; (3) the snapshot held more Taboola creatives than Outbrain, MGID and Revcontent creatives combined (108,573 + 62,765 + 15,789 = 187,127); (4) it held 1.9× as many Taboola creatives as Outbrain creatives (108,573), and the full comparison is in Taboola vs Outbrain; (5) 3.3× as many as MGID creatives (62,765), see MGID vs Taboola; (6) 13× as many as Revcontent creatives (15,789); (7) only the collection's Microsoft Audience Network label held more (281,839); (8) the creatives across all networks came from 29,257 advertisers in the collection. These ratios compare what our capture collected on each network, and capture effort differed by network, so they are not market shares and do not rank the networks. Taboola says it serves native ads on MSN, so the split between the Taboola and Microsoft labels needs re-checking before those two counts are compared.
| Network label | Captured creatives, June 2026 snapshot (collection coverage, not market size) |
|---|---|
| Microsoft Audience Network | 281,839 |
| Taboola | 206,145 |
| Outbrain | 108,573 |
| MGID | 62,765 |
| Revcontent | 15,789 |
| MediaGo | 6,571 |
| Yahoo | 5,926 |
Read the table as a description of our collection, not a ranking of networks. Capture effort, geo coverage and network attribution differ by network, so a larger count can reflect where we captured more often rather than where more advertisers spend; it says nothing about reach or auction depth, and it does not show whether Taboola leads Outbrain in the market. Microsoft's former Microsoft Audience Network product page now redirects to its Display and native ads page (checked 7 October 2026), and Taboola serves native ads on MSN, so treat the Microsoft label with particular caution. For company scale, Taboola's and Teads' own filings are the better source; the practical use of the collection is to see which advertisers, angles and landing pages appeared.
Two more index-level numbers frame everything below:
The June 2026 snapshot recorded 6,887,746 ad observations (repeated sightings of creatives across placements), an average of roughly 9.5 per creative across all networks, and 1,307,705 landing-page captures, about 1.8 per creative. The landing captures are what let us describe funnels and not just thumbnails.
Vertical labels in the June 2026 Taboola sample (statistics 11–17)#
In the June 2026 snapshot, classified Taboola creatives were concentrated in a handful of commercial verticals:
Health was the largest classified label among captured Taboola creatives: 11,982. Finance was second at 8,200 and insurance third at 7,422. Ecommerce followed at 5,185, home & garden at 4,414 and software at 3,665. The top three labels (health + finance + insurance = 27,604 creatives) held more than twice as many captured creatives as the next three combined (13,264). Health exceeded finance by 46%, the biggest gap between any adjacent pair in the top six. Health alone accounted for roughly 29% of the six largest classified labels (40,868 creatives). All of these are capture counts for the classified subset, not shares of Taboola's advertising.
| Taboola vertical label | Classified captured creatives, June 2026 snapshot |
|---|---|
| Health | 11,982 |
| Finance | 8,200 |
| Insurance | 7,422 |
| Ecommerce | 5,185 |
| Home & garden | 4,414 |
| Software | 3,665 |
These are counts of captured, classified creatives, not measures of inventory, spend or auction pressure, so they cannot show which verticals are most contested or which creative patterns work. They are useful as a map of where to look for examples. The index-wide version of this breakdown is in top native ad verticals 2026, and the advertiser-level view in who advertises on Taboola.
Why were health, finance and insurance labels so common? One hypothesis is that these offers can carry high customer lifetime value and benefit from explanation, which suits editorial-style landing pages. We have not tested that explanation, and the same snapshot also contained long-observed impulse products such as a garden ornament, so treat it as a question to test, not a finding. These are also regulated categories in which claims need substantiation and some offers need approval before they can run.
Cross-network counts in the same snapshot (statistics 18–24)#
The cross-network counts from the same June 2026 snapshot:
Taboola-labelled creatives made up roughly 49% of the classified health creatives in the snapshot (11,982 of 24,472), about 34% of classified finance creatives (8,200 of 24,068) and about 33% of classified insurance creatives (7,422 of 22,427). The Taboola health count was 3.9× the Outbrain health count (11,982 vs 3,102), the largest vertical gap between the two networks in the snapshot. The two networks' labels ranked differently: insurance was the largest classified label for Outbrain (4,345 creatives) but third for Taboola. Entertainment was the largest classified label for MGID (13,987 creatives), more than five times its next five labels combined, while Taboola's classified mix was mostly commercial. In ecommerce, the 5,185 classified Taboola creatives were 2.3× Outbrain's 2,277. These are collection counts; they do not show depth of demand, competition or which network suits a given campaign.
The practical lesson does not depend on the counts: treat each network as its own test. The ads around yours, the audience and the prices differ by network, so an angle that worked for you on one network has to be re-tested, with its own bid and device split, on the next. Our counts cannot tell you whether the receiving network will be cheaper or less competitive.
Observed runtimes and creative patterns in the June 2026 captures (statistics 25–30)#
Because the collection records when each captured creative was first and last observed, it can show how long an ad stayed visible to our capture. That is not the same as how long it ran, what it cost or whether it worked.
The longest-observed Taboola creatives in the June snapshot had first and last captures 37 days apart and were still being captured when the window closed. Two of the ten Taboola creatives in our sample set reached that mark: a home-and-garden ornament ad ("My garden had no butterflies for years — then I hung one of these up") and a hearing-device ad from Audika, one impulse product and one considered health purchase. Across all networks, the longest observed spans in the same window were 38 days, the length of the observation window itself, so the true ages may be longer. A negative-framing listicle ("Top 5 Shampoos To Avoid") had been observed for 21 days and a "surprising household item" hair-regrowth angle for 31 days; these are single examples, not evidence that the formats perform. The same "Koreans Do This Instead" skincare angle appeared on both Taboola and MGID from the same brand family, which shows an angle being reused across networks, not that it was profitable on either.
The same ten-ad Taboola sample that contained two 37-day observations also contained creatives first captured that day, with the rest spread across the range between. Ten ads are too few to describe the network, and a short observed span can mean a paused test, a budget cap, a schedule, a creative swap or a gap in our capture. Sorting by days observed is a useful way to find ads worth studying, but it does not separate winners from experiments.
Sustained runtime is not a profit-and-loss statement. An ad can stay visible for five weeks because it is profitable, but also because it belongs to a brand or awareness campaign, runs on a testing budget, is being optimised slowly or barely spends; captures show none of spend, clicks, conversions or margin. How to use runtime as a research filter within those limits is covered in our ad longevity guide, with annotated specimens in our Taboola ad examples teardown.
What these numbers mean if you buy (or watch) Taboola traffic#
For media buyers entering the network. The vertical counts show where our capture found the most examples to study, not where competition or proof is greatest; they do not validate any angle or measure auction pressure. The entry workflow is the same in every vertical: study ads observed for three weeks or more in your vertical as examples (not as proven winners), work out your own break-even CPA, and test with your own tracking. Searching and filtering the collection by vertical and days running at /spy/taboola requires the Founders plan ($29.99 a month); the free plan shows the 20 newest ads. Learn the network's mechanics in how Taboola ads work.
For advertisers choosing between networks. The cross-network counts show that our capture found different category mixes on different networks; they do not show where your offer will perform or which network has more demand. Choose the first network on eligibility under each network's policies, your own unit economics and a small controlled test, then re-test anything that worked before moving it elsewhere. The cost-benefit of the network overall is covered in is Taboola worth it.
For competitive-intelligence teams. Repeated observations and landing captures (about 9.5 sightings and 1.8 landing captures per creative across the June 2026 snapshot) are what turn a pile of thumbnails into research material: you can see where a competitor's ad was captured, when it was first and last seen, and what landing page sat behind it. They still cannot tell you which of their angles is paying.
For publishers. The vertical counts hint at the kinds of ads your readers may see in recommendation widgets, but they are capture counts, not bidder pools or demand, and they cannot tell you which content categories earn the most. Ask prospective widget partners for category and revenue data for your own site.
The June 2026 figures cannot be reproduced from the collection today: it keeps changing, and no fixed extract of that snapshot has been published. You can inspect individual captured Taboola ads, with their resolved advertiser where available, first-seen and last-seen dates, observation geos and devices, and captured landing page where available. The free plan shows the 20 newest ads; search and filters need the Founders plan ($29.99 a month).
One closing caveat: everything above is a historical snapshot of public captures from June 2026, not a current measurement. The counts describe collection coverage, not Taboola's totals or market share; vertical counts cover classified creatives only; and runtime figures are clipped by the observation window. Do not apply these numbers to later months without a fresh, reproducible extract.







