Retail media: €775M in France in H1 2026, up 18%. Before you put budget into it, look at what you are buying: captured demand, not created demand.
Retail media is the fastest-growing digital channel in France. It is also the budget line a challenger brand is most likely to misread. It gets sold to you as acquisition: a flattering ROAS, attributed sales, a dashboard that is green all over. My view is blunter. Retail media is a distribution cost first. It captures demand that already exists. It creates almost none. Budget it as a growth channel and you will pay more and more to sell to people who were already looking for you.
Retail media: the French market numbers in 2026
The 36th Observatoire de l'e-pub from SRI, produced by Oliver Wyman with UDECAM and published on July 9, 2026, sets the scale.
- €775 million in retail media revenue in the first half of 2026, or 12% of French digital advertising.
- Up 18% year on year. The digital market as a whole grew 12%.
- Retail search accounts for €599 million, up 24%. Search outside retail grew 9%.
- Retail display is stuck at €176 million, up 3%.
Do the division: 599 out of 775 means 77% of French retail media is search. In other words, sponsored listings in the search results of Amazon, Carrefour or Cdiscount. Keep that number in mind, it explains everything else.
What you are buying: shelf space, not an audience
When someone types "magnesium supplement" into Amazon, they have already decided to buy. All that is left is choosing what. A sponsored product puts your item in front of them before the next one. That is exactly what an end-of-aisle display does in a store, and nobody ever filed the end-of-aisle display under the media budget.
Kantar Media's Retail Media Search barometer, published on September 16, 2026, confirms the nature of the channel: in the first half, Sponsored Products made up 73% of the ad occurrences recorded on Amazon, against 27% for Sponsored Brands. Three ads out of four push a product page, not a brand.
That is not a flaw. Catching purchase intent in the last yard is valuable. But it sets a simple limit: search volume on your category does not depend on you, and search volume on your name depends on what you do elsewhere. Retail media harvests. It does not sow.
5,356 advertisers in the same auction
Kantar Media counted the advertisers present in Amazon search in the first half of 2026: 5,356. Over the same period, linear television had 1,845. Nearly three times fewer. The most crowded category, health supplements, lines up 252 advertisers on its own.
The number that interests me most is this one: 44% of those advertisers are absent from every other channel, offline and online. More than four brands in ten advertise there and nowhere else. They all fight over the same demand, and never manufacture any new one.
And the auction keeps rising. In its mid-year 2026 report on Europe, Pacvue measured on Amazon, for the second quarter, average daily spend up 39.6% year on year for Sponsored Brands and up 23.5% for Sponsored Products. On Amazon's DSP, return on ad spend fell 24.3% year on year. More money every year in the same auction: the price of the slot is unlikely to come down.
It is the same mechanism I describe on social platforms in the glass ceiling syndrome, with a shelf in place of the news feed.
Retail media ROAS flatters by design
The retailer sells the space, rings up the sale and writes the report. It is judge and party. The ROAS it shows counts every sale that follows a click, including the ones that would have happened without the ad. And the biggest share of those sales often comes from searches on your own name: a customer who typed your brand, whom you would have had in the organic results, and whom you pay to get back.
Advertisers know it. Skai and Stratably surveyed 166 retail media advertisers for their "2026 State of Retail Media" report: only 15% say they are very effective at measuring the channel's performance, and 75% name incrementality as their top measurement challenge. Three out of four cannot say what their budget actually added.
The answer comes down to two habits. First: look at ad spend against total revenue made at the retailer, organic sales included, not against attributed revenue alone. Second: test. Switch off the ads on your brand name for two weeks on part of the catalogue, and watch total sales. If they do not move, you were paying for nothing. If a competitor takes the slot, you will know what defence is worth. For the broader measurement question, see cookieless attribution.
My rule for a retail media budget
Three separate lines, never one. Defence: your brand keywords, at the minimum needed not to be overtaken. Conquest: category keywords, judged on new customers and not on ROAS. And a cap, set as a percentage of total revenue at the retailer, written down before launch. Anything above the cap goes elsewhere, where demand gets created.
The customer data stays with the retailer
This is the hidden cost. A customer won on a marketplace is the marketplace's customer. You have neither their email nor the right to contact them again. To sell them something else, you go back through the auction. You are renting access to your own buyers.
In the United States, eMarketer estimates that Amazon took more than 75% of retail media spending in 2025. When a single player holds the shelf, the till and the data, your bargaining power is thin.
The antidote is a base you own. For Essity, a gamified social campaign generated 14 million impressions and leads at around €6, with a first-party base as the outcome. The brand can talk to those contacts whenever it wants, with no auction. That work happens on social platforms.
Retail media for a challenger brand: go, but in the right order
Yes, if your category is searched and you are listed. Not showing up on your category's queries means leaving the shelf to others. Start with Sponsored Products on a handful of keywords, with a cap.
Yes, to defend your name, but at the right price. If a competitor buys your brand, you have no choice. If nobody does, your switch-off test will tell you.
No, as a growth engine. If nobody types your name, retail media will not fix that. It will make you pay category price for every sale, forever. A brand people search by name pays less for its clicks and converts better. And that awareness is built off the shelf: on TV and CTV, in outdoor, in social.
What I tell founders
Retail media is not optional if you sell through a retailer. But file it in the right column. It is a cost of access to the shelf, like listing fees or promotions. Cap it, measure it against total revenue, and put the growth money where people who do not know you learn your name. The brand that wins in Amazon's search engine is the one people type straight into it.
If you want to look at how your budget splits between defence, conquest and demand creation, we can do it together in a free audit. And to set your cap before launch, the ROAS calculator gives you the break-even point in two minutes.
