Retail media went from a niche line item to one of the largest categories in digital advertising within a few years. The formats involved are unremarkable — sponsored listings, banners, some video.
The growth is not about the formats. It is about a structural advantage that nothing else in digital advertising has.
The closed loop
The whole explanation in one idea.
A retailer knows what was shown, what was clicked, and what was purchased, because all three happen inside its own environment.
No cross-site tracking is required. No attribution model has to guess. The impression and the transaction are in the same database.
Which means retail media is measurable in a way that display advertising has never been, and it became more valuable precisely as everything else became less measurable.
That timing is not a coincidence. The growth of retail media and the decline of third-party tracking are the same story.
The intent advantage
The second structural factor.
Somebody browsing a retailer for a product category is closer to a purchase decision than almost any other audience an advertiser can reach.
That is the same advantage search advertising has, which is why search has always commanded premium rates, and retail media has it with the addition of transaction data.
The consequence is high conversion rates and correspondingly high rates that advertisers will pay.
The margin story for retailers
Why retailers pursued it so aggressively, which is worth understanding as an advertiser.
Retail is a low margin business. Advertising is a high margin business.
A retailer selling advertising against its own traffic is adding revenue with very little incremental cost, and that revenue drops largely to the bottom line.
For several large retailers, the advertising business now contributes a share of operating profit disproportionate to its share of revenue, which is a strong incentive to expand it.
This is also why the inventory keeps growing — more sponsored slots per page, more placements, more formats — which is the pressure that eventually degrades the customer experience.
The incrementality question
The serious analytical objection and the one advertisers should be pressing on.
A sponsored listing shown to somebody who searched for your product on a retailer's site converts well. The question is whether they would have bought it anyway.
For a branded search on a retailer where you are already the leading result, the honest answer is frequently that a substantial share of those conversions were going to happen.
Retail media reporting, being a closed loop, is very good at attributing and no better than anything else at establishing incrementality.
Advertisers who have run proper holdout tests in this channel report results that vary enormously by category and by product, and the tests are not always welcomed by the platform.
The standardisation problem
The practical difficulty of operating in the channel.
Every retailer runs its own platform, with its own metrics, its own attribution rules, its own reporting formats and its own definitions of a conversion.
Comparing performance between two retailers is genuinely difficult, and aggregating across five is a manual exercise.
Industry bodies have published standards for this and adoption is partial, because each retailer's numbers currently look better under its own methodology than under a common one.
The practical response is to insist on the raw data where possible and to normalise yourself, which is work and which is the only way to compare fairly.
Where it goes next
Two developments worth watching.
Off-site extension, where retailers use their purchase data to target advertising on other properties. That takes the data advantage outside the closed loop and reintroduces all the attribution problems, while retaining the targeting advantage.
And the extension of the model beyond retail. Any business with logged-in users and transaction data has the same structural advantage — travel, financial services, delivery platforms, healthcare in some markets. Several have already launched advertising businesses on exactly this logic.
What an advertiser should actually do
Treat the reported returns as a ceiling rather than a measurement, and test incrementality where the spend justifies it.
Negotiate for data access rather than accepting the standard reporting, which is increasingly possible at scale.
Normalise metrics across retailers yourself rather than comparing their dashboards.
And keep an eye on total sponsored density on the retailers you use, because the customer experience degradation eventually affects the value of the placement, and it is happening.
The in-store extension
The development that is furthest along and least discussed outside the sector.
Screens at shelf, at checkout and on trolleys, sold programmatically, measured against till data.
The measurement here is genuinely close to a closed loop, since exposure and purchase happen metres and minutes apart.
The constraint is that attribution to an individual requires loyalty card identification, which is available for a proportion of shoppers and not all.
For advertisers it is worth watching because the inventory is growing quickly and the rate cards are still being established, which historically is when the value is best.