Open in interactive viewer → charts, metric popovers & call review

Criteo’s AI Story Is Still a Test Budget

Pipeline is up 72% and CTV is doubling, but revenue is shrinking, retail media is down 23%, and 2027 remains a promise.
CRTO · Earnings Call · 2026-11-04

A beat that doesn’t change the trajectory

Criteo’s third-quarter report lands in the middle of a de-rating, not at the start of one. The stock is down 55.8% over its full history window and 14.4% in the last 90 days; it sits 32.8% below its July 13 peak of $23.28. At a $766 million market cap, the market is already treating the advertising agency as a melting legacy asset. The quarter did little to break that framing: revenue was $416 million, Contribution ex-TAC $240 million, down 15% at constant currency, and adjusted EBITDA $65 million. The EBITDA beat came from lower hiring and a partial reversal of accrued bonus expense, not from a demand snapback. “our top line was in line with our expectations, and Adjusted EBITDA was ahead of plan” — Connor McGogney, Chief Financial Officer · 2026-11-04, CFO Connor McGogney said. That is a cost story wearing a beat’s clothes. The last filed quarter showed the underlying pressure: total revenue of $428 million, down 23% year over year, with operating margin at 3.6%, down 13.9 percentage points. Management kept full-year guidance unchanged: Contribution ex-TAC down 10%–12% at constant currency, adjusted EBITDA margin around 30%. That unchanged guide matters. It means Q3’s cost beat did not buy a better revenue outlook. “we have not assumed an immediate benefit in our remaining 2026 outlook” — Connor McGogney, Chief Financial Officer · 2026-11-04, McGogney said. Investors are left with a familiar Criteo pattern: credible expense discipline, weak top-line conviction, and a 2027 recovery story that has not yet shown up in reported numbers. The problem is not new. In August, then-CFO Sarah Glickman described “a number of, I would say, extra large U.S. clients, and they’re all kind of down” — Sarah Glickman, Chief Financial Officer · 2026-05-06.

New business is the new headline

The freshest company-specific keyword in the quarter is New business. Criteo says open opportunities are up 72% year over year, agencies represent more than half the pipeline, new-business spend grew 13%, and the share of new enterprise client spend doubled. That is a real change in commercial activity, and management deserves credit for at least measuring the funnel more aggressively. The strategic bet is that Criteo can sell more than legacy retargeting: CTV, social, apps, and AI platforms now represent 13% of Commerce Growth campaigns, up from 6% a year ago. CTV is the clearest proof point. Criteo is expanding the offering globally after early U.S. success, with Brazil and Mexico by year-end and major European markets, Canada, and Australia in 2027. Management says it is on track to double the Q3 CTV run rate in Q4, citing a travel platform that generated a 61x return on ad spend and 9% lower cost per acquisition than display. The differentiator is the commerce graph: Criteo can build audiences from actual shopping behavior rather than proxy signals. That is a plausible right-to-win in performance CTV, where Walmart’s Vibe.co acquisition and Amazon’s Prime Video are already competing for the same budgets. The question is scale. CTV may be doubling off a small base. The same is true of ChatGPT Ads. Criteo’s OpenAI integration went from seven countries last quarter to 89, and 3,600 clients have been on and off the platform. The Turkish Airlines example — a 183% increase in site landings and 15% more time per visit — is a good demo. But Michael Komasinski is explicit: “This is still an early market, and spending has yet to scale” — Michael Komasinski, Chief Executive Officer · 2026-11-04. Clients test, leave, and return; average spend per client is rising only modestly. ChatGPT Ads is a promising new channel, but it is not yet a revenue line that can offset the legacy client losses.

Retail media’s scope cuts are still the story

Retail media is where the report gets uncomfortable. Revenue was $52 million and Contribution ex-TAC $51 million, down 23% at constant currency, including a $20 million hit from the previously communicated client scope reductions at two clients. Excluding those cuts, underlying retail media Contribution ex-TAC grew 10% — healthy in isolation, but a deceleration from the 20% growth Criteo reported in Q2. Management points to tougher comparisons, delays at larger retailers pushing display launches into 2027, and a slower October start. The market is maturing; Criteo is still growing, but no longer at the rate that supported the old multiple. The product story is better than the reported number. Auction-based display is live with more than 95 retailers, twice the start of the year, and accounts for more than half of on-site display; display spend is on track to grow close to 50% this year. Four of Criteo’s largest retail partners renewed multi-year agreements, and new partners such as Notino, El Puerto de Liverpool, ONIGO, Adore Beauty, and PayU broaden the network. But display carries a lower take rate than classic on-site sponsored ads, and the mix shift shows up in gross margin: 51.9%, down 2.5 percentage points year over year. More volume, less yield per dollar — a familiar ad-tech trade-off. The existential debate is Agentic commerce. Analysts pressed management on whether AI assistants disintermediate retail media by taking the purchase decision away from retailer sites. Komasinski’s answer is the most interesting strategic argument in the call: horizontal agents like ChatGPT handle discovery, but retailer-owned vertical agents complete basket building, payments, fulfillment, and service.

We see sort of the world similarly to what other executives, I think, have talked about with horizontal agents and vertical agents... retailers continue to have a vital role in the shopping experience.

Michael Komasinski, Chief Executive Officer · 2026-11-04
Criteo’s play is to bring sponsored recommendations into those retailer-owned AI assistants, as it is doing with Albertsons and THG Ingenuity, and to use Page Intelligence to optimize product discovery on retailer sites. It is a credible defense, but it is still early. The AI-assistant theme is everywhere this earnings season; Criteo’s version is company-specific only if retailers actually keep control of the transaction.

Cheap for a reason, with optionality

The balance sheet is the strongest part of the story. Criteo ended September with $275 million in cash and no long-term debt, and effective net cash of $275 million is a real cushion. It repurchased $4 million of shares in Q3, bringing year-to-date buybacks to $65 million, with $156 million remaining. The planned redomiciliation to the U.S. on January 1, 2027 could broaden the investor base and create index inclusion potential. Those are supports, not growth drivers. The valuation already discounts a lot: 0.5x price to revenue, down 58% year over year. But cheap ad-tech can stay cheap when the core revenue line is shrinking. The bull case requires new business to ramp in the first half of 2027, retail media to lap the scope cuts, and AI/CTV to become material rather than experimental. The bear case is that Criteo is using AI language to describe a business that is still losing large clients faster than it can replace them. Prior calls already carried the warning: “we need to have our new business engine and our new products performing at a level that overcomes those ups and downs” — Michael Komasinski, Chief Executive Officer · 2026-08-05. That was August. It is still the central task. The quarter’s real signal is not the EBITDA beat. It is that management’s language has shifted from defending large-client losses to selling a pipeline — while the reported numbers remain negative. “This has been a challenging year for Criteo. Our performance is not yet where I want it to be” — Michael Komasinski, Chief Executive Officer · 2026-11-04, Komasinski said. That is honest. The stock is already priced for worse than honest. What it needs now is evidence that the new channels and new business are large enough to matter before the legacy base shrinks further.