Retail media: the invoice nobody audits
Retail media is the fastest-growing line in most consumer marketing budgets and the least audited. The reported returns arrive at six, eight, twelve times spend, and nobody wants to ask the uncomfortable question: how much of that would have happened anyway?
The short version
- Retail media harvests demand, it does not generate it. Targeting people already in the store, already in the category, reports presence at conversion as cause of conversion.
- Consistency is the tell. A channel returning the same high multiple every quarter regardless of creative is counting, not performing.
- Only incremental return belongs in a P&L - a campaign reporting eight times commonly lands between one and two times on a clean holdout.
- Watch for the double count: retail media funded from trade terms can appear both as a deduction from net revenue and as media investment, in systems owned by different directors.
- It is structurally poor at growth, because it cannot reach the light and non-buyers who are not in the store thinking about you.
Retail media is the fastest-growing line in most consumer marketing budgets and the least audited. The pitch is irresistible: closed-loop measurement, first-party data, an ad served metres from the purchase. The reported returns arrive at six, eight, twelve times spend, and nobody in the room wants to be the person who asks the uncomfortable question. So here it is. How much of that would have happened anyway?
Why retail media always looks brilliant
A retailer's network sells you attention among people who are already in the store, already in the category, and frequently already holding your product. Targeting that precise is not a demand generator; it is a demand harvester. Attribution then credits the last touch before a conversion that was largely determined before the ad was served. The mechanism is not dishonest, it is simply measuring the wrong thing: presence at the moment of conversion is being reported as cause of conversion.
The tell is the consistency of the numbers. Genuine media effects vary wildly by creative, category and season. A channel that returns comfortably high multiples every quarter regardless of what you put in it is not performing; it is counting.
The only number that belongs in a P&L
Incremental return requires a holdout: a matched set of stores, regions or audiences deliberately excluded, measured against the exposed set. It is operationally annoying, the retailer rarely offers it unprompted, and it is the difference between a media investment and an accounting exercise. In the tests we have seen, a campaign reporting eight times return commonly lands between one and two times on a clean holdout. That is not a scandal - one point four times can still be a perfectly good trade. It is only a scandal if you built the budget on eight.
A channel that returns the same high multiple every quarter regardless of the creative is not performing. It is counting.
The double count nobody wants to find
Here is the structural problem underneath the measurement problem. Retail media is frequently funded out of trade terms rather than the media budget - negotiated into the annual agreement alongside listing fees, promotional support and rebates. When that happens, the same euro can appear twice in the company's own reporting: as a deduction from net revenue in the commercial P&L, and as media investment in the marketing dashboard. Marketing efficiency improves on paper while gross margin quietly erodes, and because the two numbers live in different systems owned by different directors, the contradiction can run for years without anyone reconciling it.
This is the same waterfall problem we described in the discount nobody costed, arriving through a new door. The question to put to the finance team is blunt: show me retail media spend in the trade accrual and in the media plan, and tell me whether any of it is the same money.
What retail media is genuinely good at
None of this means the channel is worthless. It is unusually good at three jobs, and they are worth paying for honestly. Defending a listing when a competitor is on promotion. Buying visibility for a genuinely new SKU that has no baseline demand to harvest, where almost all measured sales are by definition incremental. And converting shoppers at the moment a substitute is being considered, which is a real intervention rather than a claimed one. Fund those, measure them against a holdout, and the channel earns its place.
What it is bad at, and why that matters more
Retail media reaches people already buying the category, which makes it structurally poor at the one thing that grows brands: bringing in light and non-buyers who are not in the store thinking about you. A budget that migrates steadily from broad reach into retail media will show improving efficiency and declining penetration at the same time, and the efficiency metrics will be cited as evidence that the strategy is working. We set out why that trade is usually wrong in penetration, not loyalty.
How we cost it
This is Cashstream work: retail media pulled out of both the trade accrual and the media plan, reconciled into a single line, and restated on incremental return rather than reported return. Shopperology supplies the shopper-side read on where in the journey the spend actually intervenes, which decides whether a format is harvesting or converting. The output is a budget you can defend in front of a CFO and a commercial director in the same meeting, which is a harder test than either one alone. Our strategic advisory page explains how this fits a wider engagement.
Suspect your retail media returns are harvesting rather than working? We will rebuild the number on a holdout basis and find out whether any of it is trade money counted twice.
Book a call