Taboola's Page Views Didn't Fall. The Scrolling Did.
A guidance cut built on a brand-new mechanism — fewer ad impressions per page view — lands on a stock already down 45% in ninety days and buying back its own shares into the hole.
TBLA · Earnings Call · 2026-11-04
The only number that changed is the one nobody was watching
For four straight quarters Taboola.com sold investors a single, tidy story: the open web was fine, AI exposure was single-digit, and Realize would deliver a return to consistent double-digit growth. On November 4 the company cut its full-year ex-TAC guide to $758 million — roughly 6% growth, down from the 7% it had been guiding — and the reason had almost nothing to do with the things management had spent a year reassuring people about.
It wasn't referral traffic. CFO Steve Walker was blunt when Laura Martin of Needham asked directly whether a Google search-algorithm change was the culprit: “The simple answer to your question is no, it's not less referral traffic from Google that we're seeing. It's really about the fewer impressions per page view.” — Steve Walker, Chief Financial Officer · 2026-11-04 Page views, the top of the funnel everyone feared AI would gut, held roughly steady. What broke was the middle of the funnel — the number of ad slots actually rendered once a reader arrived.
This is a genuinely new keyword regime for the company. page views rocketed to the number one slot on Taboola's keyword grid this quarter, accompanied by impression volume headwinds and a plain-language diagnosis of shifting consumer behavior. Compare that to the last four quarters, which were about Fox News wins, network cleanup, minimum guarantees and Realize+ adoption. The vocabulary of the call changed because the business did.
The margin paradox hiding inside the revenue decline
Here is the part that makes this quarter more interesting than a simple miss. Q3 revenue fell 7% year over year to $462.3 million, yet ex-TAC gross profit rose 7% to $188.8 million, gross profit rose 10%, and adjusted EBITDA rose 16% to $56 million at a 29.7% margin. The mechanism, per Walker: fewer impressions mean the same advertiser demand gets spread over less supply, which pushes up higher ad rates and therefore higher take rate.
That is a real but treacherous trade. A company can shrink its way to better margins for a while; it cannot do so indefinitely, because the volume that funds the margin is the same volume walking out the door. The full-year guide now implies revenue of $1.88–1.90 billion and ex-TAC of $754–763 million against adjusted EBITDA of $212–221 million — structurally healthy profitability on a shrinking top line. The most recent filed quarter printed $477 million of revenue, up just 2% year over year; the call reveals the next one going backwards.
What management stopped saying
The loudest signal this quarter is an absence. Scaled advertisers — the metric Walker has called the company's growth fuel for years — declined 4% year over year, and average revenue per scaled advertiser fell 5%. These are the roughly 85%-of-revenue customers whose growth Taboola has marketed as proof the platform works. The keyword number of scaled advertisers is one of only a handful of true decliners in the company's keyword history; this quarter it stopped being a talking point altogether.
That reframes the prior calls. In February, Walker said of the growth trajectory: “We still feel like we've seen an inflection in our business... I wouldn't say anything has changed on the time line.” — Stephen Walker, CFO · 2026-05-06 In August, Adam Singolda closed by noting “we raised our guidance, again” — Adam Singolda, Founder and CEO · 2026-08-05 — after buying back approximately 20% of the shares since 2025. Today the only forward-looking statement about growth is Walker's: “our ambition as a management team is to get back to double-digit growth” — Steve Walker, Chief Financial Officer · 2026-11-04 — ambition, not trajectory.
The other dropped thread is search exposure. In August 2025 Singolda explained that “we have seen a minimal impact to date from LLM-driven search changes... about 5% of our U.S. traffic as of now comes from search” — Adam Singolda, Founder and CEO · 2025-08-06, and argued that publishers would birth a new, more valuable traffic type on their own sites. That framing was technically correct and commercially beside the point: the damage came through engagement depth, not referral volume. Barton Crockett of Rosenblatt pressed on breadth, and Singolda conceded it is “kind of across the board” — Adam Singolda, Founder and Chief Executive Officer · 2026-11-04, adding that it is hard to reverse-engineer why readers behave differently — they expect a summary at the top rather than scrolling to the bottom.
The offsets: publishers, Dianomi, and the LLM bet
Management's answer is three-pronged and each prong has a keyword. First, win supply: the Fox News, Fox Local, Fox Sports and Fox Weather signing (FOX News) plus a claimed new publisher pipeline. Second, monetize deeper: Full Page Monetization is now live across 100% of ad inventory at NBC News and TODAY.com, which management believes can lift revenue per publisher two- to threefold. Third, buy supply: the proposed acquisition of Dianomi, which brings hundreds of premium finance and lifestyle publishers — Reuters, CNN Business, The Wall Street Journal — expected to close before year-end 2026.
And the hedge: DeeperDive, the AI answer engine now at 100-plus sites and roughly 14 million users, plus a Motorola device integration. Singolda claims Taboola is “one of the only companies that I'm aware of that actually is showing ads in an LLM conversational layer” — Adam Singolda, Founder and Chief Executive Officer · 2026-11-04, and that a conversational layer will migrate onto apps, TVs and any digital canvas. It is the most interesting thing the company owns, and it is explicitly not yet material: “if this was a startup, it would be a very, very fast-growing startup” — Adam Singolda, Founder and Chief Executive Officer · 2026-11-04 is the CEO's own framing of immateriality.
The capital-allocation question nobody asked
Taboola repurchased about 7 million shares in Q3 at an average of $4.37 for $30.3 million, holds $90 million of remaining authorization, and just added $200 million more. It ended the quarter with $50.2 million of net cash and roughly $191 million available under a $270 million revolver. Meanwhile the stock sits 50.9% below its July 9 peak of $5.58 and 45.2% lower over the last 90 days, leaving a market capitalization near $919 million. Quarterly buybacks peaked above $100 million in mid-2025 and have been running $42 million a quarter, against a price-to-revenue multiple of 0.9x.
Buying a shrinking business at a low multiple can be rational; doing it while the core impression base erodes and the AI thesis is still a startup inside a startup is a leap of faith. Sharing the tape's suspicion: the broader market is also de-rating generative AI capability stories that have not yet converted into revenue, which is precisely the box DeeperDive sits in.
The bottom line
The honest read is that Taboola has discovered the second-order effect of AI summarization before most of its peers could name it, and it discovered it the hard way: not in traffic, but in the number of things a reader is willing to look at before leaving. Page views flat, impressions down, ad rates up, revenue down 7%, ex-TAC up 7%. That combination is coherent, and it is not a one-quarter event — Walker said the signs appeared earlier this year and were masked by the Explore More deprecation and network cleanup.
He also said the Q4 guide can be used as an indication of the 2027 base case. That is the number to hold him to in February. If Taboola News, Full Page Monetization and Dianomi can grow supply faster than engagement per page erodes, the margin-led story survives. If not, the open web's self-appointed monetization infrastructure is simply a ad tech name with a lean cost base, a collapsing share count, and a shrinking market.