Criteo's Profit Fell 85% — While Advertisers Spent 9% More

Criteo's profit fell 85% in one quarter, from $40 million to $6 million. Revenue fell 11%. The part that matters: the media spend flowing through its platforms rose 9% at constant currency anyway.

Criteo's Profit Fell 85% While Ad Money Through It Rose

Criteo's third-quarter results landed on November 4, 2026, and the top number is ugly. Revenue for the quarter ended September 30 fell 11% to $416 million, down from $470 million a year earlier.

Net income dropped 85%, landing at $6 million, down from $40 million. Diluted earnings per share were $0.11, down from $0.70 a year earlier.

Contribution ex-TAC — the number Criteo steers by — fell 17% to $240 million. Adjusted EBITDA fell 39% to $65 million, and the margin on that measure narrowed to 27% from 36%.

Here is the number that matters. The media spend flowing through Criteo's platforms climbed 9% at constant currency, reaching $1.2 billion. Advertisers pushed more money through the machine while the machine kept a smaller slice of it.

"This has been a challenging year for Criteo," chief executive Michael Komasinski said on the call. He said the results are not where he wants them yet, and that what the company can do and the progress it is making leave him confident about where it is headed.

The client count tells its own story. A table in the release puts the count at 16,703 at the end of September, 2% below a year earlier and the weakest of the nine quarters it reports. Headcount fell 4% to 3,504.

Criteo is trimming costs, reworking how new clients sign up for its self-service tool, and placing bets on newer ad surfaces such as streaming TV and ChatGPT. Komasinski said the company is phasing out standalone self-registration and using automated tools to handle the first steps of onboarding.

Its newest bet is barely a bet yet. Criteo joined OpenAI's ChatGPT advertising pilot in March 2026 as its first ad tech partner, and Komasinski said the offering is now live in 89 countries, up from seven countries on the previous call. His words on the money: "This is still an early market, and spending has yet to scale."

He expects ChatGPT Ads to start contributing in 2027.

So a company that sells attention watched its profit fall off a cliff while advertiser money through it grew, and it is reworking how customers sign up. That is not a one-company story.

Snap Changed the Number It Reports

Snap shares fell over 10% after the company missed on both revenue and daily users in its third-quarter report.

Sales still jumped 14% year over year, and net income reached $56 million, compared with a $104 million loss a year earlier.

Then it changed the number it reports. Snap told investors it will make worldwide monthly users its main audience number starting in the first quarter of 2027, according to its investor letter.

Snap's investor letter said daily use has held a steady share of monthly use over time, and that the company will not trade long-term value for its users to chase daily visits.

Snap's commercial chief, Ronan Harris, told analysts that lower spending by unnamed consumer packaged goods companies hit third-quarter sales, and he tied it to broad economic pressure in the U.S. and the Middle East. He said the company is seeing encouraging growth in other verticals, such as U.S. financial services.

Snap is switching the number it reports, and Criteo's main profit measure kept shrinking. Now look at where the actual profit showed up in the same stretch. It was not where most people would look.

Rivian Made Its Money on Software, Not Sheet Metal

Rivian topped Wall Street's third-quarter earnings estimates while it pushed output of its new R2 SUV higher.

The split inside its numbers is the story. Its software and services segment generated $605 million in revenue and $256 million in gross profit — a 42% margin. Its automotive operations booked close to $1.55 billion in revenue and lost $80 million at the gross line.

The R2, whose first deliveries arrived in June, became Rivian's best-selling model and drove much of the gain from the second quarter. Rivian delivered 19,248 vehicles from July through September, a company record.

It also raised its full-year delivery guidance to between 67,000 and 70,000 units from at least 65,000, and said it expects the R2 to turn a positive gross margin by year end. CEO RJ Scaringe told CNBC's Phil LeBeau the ramp-up is critical for getting there, and that unit economics will keep growing into the first quarter of next year.

So here is the stretch in one line: Criteo's profit fell 85%, Snap is switching the number it reports, and a week earlier the car company earned its margin on software while the vehicles ran a gross loss. Now let me tell you how Criteo got here.

Criteo Started 2026 Expecting Growth

Criteo began 2026 guiding to flat to 2% growth. Two cuts later, it is guiding to a decline of 10% to 12% in its main profit measure for the year.

The cause sits in its biggest customers. Criteo's finance chief said budgets from several big enterprise accounts came in below the year before, and Komasinski described the job plainly when an analyst asked: "stabilizing the leaky bucket with the large clients."

Geography did not help. Criteo's CFO said the U.S. remains the most challenged market, while EMEA and Asia-Pacific were comparatively stronger. Travel, one of its verticals, decelerated, and retail stayed soft, with spending in discretionary categories such as fashion down 14%.

Costs are coming down to match. Costs booked for restructuring, integration and transformation reached $25.5 million across the first three quarters, up from $9.3 million a year earlier, and the finance chief said the company is taking a proactive look at its cost structure.

Put simply: the company that puts ads in front of shoppers got smaller twice in one year, and it is trimming to stay profitable. If you buy ads, that is worth ten minutes of your attention. Here is what I would do with it.

What This Means if You Sell Roofs or Cars

(From here down it is my read, not reporting.)

Picture a roofing contractor in a mid-size market. Facebook and Instagram send him a steady drip of form fills every week. He counts the form fills, pays the bill, and moves on.

Then a storm hits two towns over, and he has no idea whether his ads ever ran there, because his targeting is one circle drawn around his shop. Meanwhile, the middlemen who sell ad space are cutting costs and changing how customers sign up.

Nothing about the price of a roof changes because Criteo's profit fell 85%. What changes is the scoreboard he is trusting. When the people selling attention start swapping the metrics they report — Snap just told investors it is switching its primary metric — the only number left standing is the one in your own books. Cost per inspection booked. Cost per signed roof.

Picture a car dealer pushing used trucks into a weekend sale. Same logic, different math. The click is not the sale, and the Rivian split shows why that matters: $256 million in gross profit came from software and services while the vehicles themselves ran an $80 million gross loss. Its smaller, less expensive model is the one that took off.

My read for a dealer: the traffic you buy is the front door. The profit is attached to the back end, and the unit that moves volume is usually the one with the smaller sticker. If you buy traffic and never trace it to the finance office, you are measuring the least profitable half of the deal.

One more thing. Criteo says spending on ChatGPT ads has yet to scale, and it expects the contribution to build in 2027. My read: do not move this year's budget just to prove you were early. Keep the money where you can trace it to a closed job.

Here is what I would actually do about it this week.

Seven Things to Do This Week

1. Pull 90 days of ad spend and write down what the platform says you got. Leads, calls, clicks — raw numbers, on paper. No screenshots, no dashboards.

1. Count the closes for the same 90 days from your own records. Booking sheet, CRM, calendar, sales log. The gap between those two lists is either your problem or your win.

1. Put a tracking number on every ad. Then a call can be matched to a campaign instead of guessed at. Ask your phone provider to set up a tracking number for the ad line.

1. Fix your towns. Write down the towns you actually serve, by name, then open your ad targeting and check it matches. If it says a 25-mile radius around the shop, change it to town by town.

1. Build a one-page Friday sheet. Week, spend, leads, calls, appointments booked, jobs sold, revenue. Fill it in every Friday with a coffee. If you want a template, ask an AI tool for a simple weekly ad tracking sheet with those columns, short enough to fill in by hand.

1. Set one number you will act on and write it on the wall. Your cost per closed job. If it climbs two months running, cut the weakest campaign, not the whole budget.

1. Write two new ads this week, not one. Two opening lines, two images, same offer. You learn faster when they run side by side. Ask an AI tool for three Facebook ad angles for a used truck weekend at your dealership — each with a hook line, one reason to come in and one clear next step, in plain language aimed at nearby towns. For the roofer, ask for a two-sentence follow-up text to a homeowner who filled out a free inspection form three days ago and has not booked — friendly, no pressure, one clear ask.

That is a week of work, and it will tell you more than any earnings call will.

Mistakes That Cost You Money

• Judging ads on clicks and reach. Reach has never paid a supplier. Cost per closed job does.
• Chasing the newest channel with this year's money. Criteo's own numbers say spending on ChatGPT ads has yet to scale, with the contribution expected in 2027.
• Trusting one platform's headline number. Snap just told investors it is switching to a different primary metric. If they can swap scoreboards, anyone can.
• Buying a blob instead of towns. If you cannot name the three towns your money reached last month, you are guessing.