What each format actually is
On-site retail media covers any paid placement on a retailer's own digital properties: sponsored product listings on a grocery app, banner ads on a retailer's website, and display slots that appear while a shopper is browsing or searching. The retailer controls the inventory, sets the CPM, and owns the transaction data. Because the shopper's session from ad impression to purchase happens inside a single platform, the retailer can at least attempt to show you a before-and-after sales picture.
In-store retail media covers physical placements inside the store: digital screens at the end of aisles, shelf-edge LED strips, audio spots over the store PA, and screens mounted above checkout lanes. A shopper walking the snacks aisle sees your brand's ad on a screen two metres away and then reaches for the product. The ad and the purchase are close in time and space, but the data systems connecting them are rarely joined up in the same way.
Both formats sit inside what the industry calls retail media networks, and both are growing fast. Bain reports that brands are now directing 39 percent of their ad spend through retail media networks, a figure that underscores how central these channels have become to commercial planning even as measurement lags behind spend.
Measurement maturity
On-site retail media has a meaningful head start on measurement. Because the retailer's first-party data covers both the ad impression and the purchase, you can build a sales-lift study using the retailer's own transaction records. The methodology is not perfect, but the inputs exist. You can segment exposed versus unexposed shoppers within the same platform and compare basket outcomes.
In-store is harder. A shopper who sees your ad on an end-of-aisle screen is not logged in. They are not identifiable in the same way a shopper who clicks a sponsored product is. Linking screen impressions to individual transactions requires either loyalty card matching (which covers only the share of shoppers who scan their card) or panel-based estimates that introduce statistical noise. Neither approach is anywhere near as clean as a closed digital loop.
The honest assessment: neither format has solved incrementality. The industry still cannot reliably separate sales that retail media caused from sales that would have happened anyway, as the structural gap between spend and proof persists across both formats. On-site is ahead, but it is ahead by degree, not by kind.
CPM economics
On-site retail media CPMs are high and rising. Retailers know their first-party data is valuable, and they price accordingly. Sponsored search on a major grocery platform can command CPMs several times higher than a comparable open-web display placement. For a brand with strong organic search rank on the retailer's site, this means you may be paying to protect sales you already had.
In-store CPMs are structured differently because the inventory is physical. A retailer sells screen time in loops, often by store cluster, daypart, or campaign flight. The cost per thousand impressions can look attractive because in-store foot traffic is large, but the denominator (shoppers who actually see your specific ad at the specific moment) is harder to verify than a logged digital impression. Reach is high. Verifiable, audited reach is lower.
For brands trying to compare the two on a cost-per-verified-impression basis, on-site wins on transparency. For brands thinking about cost-per-shopper-at-the-shelf-moment, in-store makes a credible case, especially in high-impulse categories where proximity to purchase is the whole point.
Where each format earns its place
On-site retail media is strongest when your goal is to intercept a shopper who is already searching or browsing with buying intent. A sponsored listing for your protein bar when a shopper types "high protein snacks" into a grocery app is a very different asset from a banner ad shown to someone who is not yet in a purchase mindset. The closer the shopper is to a decision, the more value the on-site placement has.
In-store retail media earns its place in categories where the purchase decision is genuinely made in the aisle rather than at home. Roughly half of grocery purchase decisions are made inside the store according to longstanding shopper marketing research, though that share varies sharply by category. For impulse buys, new product launches where awareness is low, and categories where shoppers browse before choosing, an end-of-aisle screen or shelf-edge display can shift consideration at the moment it matters.
In-store also reaches shoppers who never interact with the retailer's app or website. A significant share of grocery shoppers, particularly older demographics, buy in store but do not shop online with the same retailer. On-site retail media misses them entirely.
Practical allocation thinking
A useful starting frame: on-site retail media is a performance channel that should be judged on sales lift against a holdout group, with CPM as a secondary efficiency metric. In-store retail media is closer to a brand and awareness channel that happens to operate near the point of purchase, and it should be judged partly on reach and partly on category-level sales trends during the flight.
Running both with no measurement discipline is how budgets leak. Running neither is how you cede shelf presence to a competitor who is willing to spend. The answer is not to exit retail media. It is to fund the placements where you can build a credible measurement case first, and treat the rest as awareness investment with explicit caps.