Where the money is going
Retail media has moved from an experimental budget line to a structural commitment. Bain projects the global retail media market will reach roughly $140 billion by 2026, a scale that makes it one of the largest advertising categories in the world. CPG companies are already directing 39 percent of their total ad spend through retail media networks, and the expectation across the industry is that share will rise to 50 percent.
That shift is not happening because brands have suddenly fallen in love with sponsored search placements. It is happening because the shopper data that retail media networks sit on top of is the closest thing to a guaranteed audience that brand budgets can buy. When a retailer knows which shoppers buy your category, which ones lapse to private label at a specific price point, and which ones are in an active replenishment cycle, the ad inventory attached to that data looks very different from a generic display placement.
The practical consequence for commercial leaders is that the planning conversation with your top retail partners has changed shape. Trade spend and media spend are converging in the same joint business plan, often controlled by the same retail partner team. If your organisation still manages those budgets in separate silos, you are almost certainly duplicating reach and missing the optimisation that happens when both pools are planned together.
The Amazon weight problem
Amazon's grocery business reported more than $150 billion in grocery sales in 2025, a figure that puts it in the conversation for second-largest U.S. grocer by sales. That commercial weight sits directly underneath Amazon's retail media network. When a brand buys a sponsored placement on Amazon, it is buying access to a shopper population that already has a payment method stored, a delivery address confirmed, and a purchase history Amazon can model against.
That combination of scale and first-party data is what makes Amazon's network structurally different from most other retail media offerings. But it also means that brands without a strong Amazon content and pricing strategy are funding a media network that is, in practice, surfacing their competitors to their own likely buyers.
The gap between Amazon and the next tier of retail media networks is still wide. Grocery banners such as Kroger, Albertsons, and Walmart have invested heavily in their own networks, but the audience data they can activate, and the measurement they can offer against it, varies considerably. For a category director building a network plan across multiple retailers, that variance matters: you cannot apply the same measurement expectations to every placement you buy.
On-site, off-site, and in-store: three different measurement problems
Retail media spending now splits across three distinct environments, each with a different measurement challenge.
On-site placements, sponsored search and display within a retailer's own app or website, are the easiest to attribute. The shopper searched, saw a placement, and either added to cart or did not. The problem is that a large share of shoppers who would have bought your product anyway are exposed to these placements, so raw conversion numbers overstate the incremental lift. Without a holdout test or a clean control group, your cost-per-acquisition looks better than it is.
Off-site placements take the retailer's first-party audience data and activate it against programmatic inventory elsewhere: connected TV, social feeds, display networks. Attribution here is harder because the purchase event happens in a different environment from the ad exposure. Retailers are building their own clean-room infrastructure to close that loop, but the methodology is not yet standardised across networks.
In-store digital media, screens at shelf, checkout lane placements, digital end-caps, is the fastest-growing format and the one with the least mature measurement. Foot traffic and dwell time can be proxied, but translating those signals into a clean sales-lift number at the SKU level is still a work in progress for most networks. The brands willing to run structured pilots with in-store formats now are building the measurement playbooks that will matter when budgets here scale.
What the incrementality gap actually costs you
The structural risk in retail media is not that you spend too much. It is that you spend on the wrong mix because you cannot see which placements are generating new sales versus capturing sales that were coming anyway.
A brand running sponsored search on its core SKUs during a period of strong category demand will see impressive attributed sales. But if category shoppers were going to buy regardless, all you have done is pay to show up in a search result that was already yours. The incrementality test, running a randomised holdout where a matched group of shoppers does not see your placement and comparing outcomes, is the only reliable way to separate the two effects.
Most CPG brands are not running those tests consistently across their retail media buys. The reasons are partly logistical (retailer cooperation is uneven), partly organisational (the team buying the media is often not the team that can run a clean measurement study), and partly a collective action problem: if everyone is measuring the same way and using the same flawed metrics, no single brand has an obvious incentive to challenge the numbers publicly.
The consequence is that budget allocation decisions inside retail media plans are often driven by attributed ROAS figures that are materially overstated. Brands that build a rigorous incrementality discipline now, even imperfect, are better positioned for the moment when the industry standardises on cleaner metrics and the networks that cannot deliver real lift become visible.
What retailers are building to close the gap
Retailers know the measurement problem is real and are investing to address it, partly because their long-term ability to command premium CPM rates depends on proving the networks work.
Kroger's 84.51 unit, Walmart Connect, and Amazon's advertising console have each expanded their measurement offerings in the past 18 months. Clean-room partnerships with third parties such as Circana and NielsenIQ allow brands to match retailer transaction data against their own audience data without either party exposing raw records. That infrastructure is becoming the baseline expectation for any serious retail media partnership.
The direction of travel is toward closed-loop measurement: a brand buys a placement, a shopper is exposed, the retailer's loyalty data records whether a purchase followed among exposed versus unexposed shoppers, and the result feeds back into planning. That loop works well when the purchase happens in the same retailer's ecosystem. It becomes harder when the shopper sees an off-site ad on connected TV but buys in a physical store belonging to a different banner.
For brands, the practical implication is that your retail media measurement capability is now a negotiating asset. Retailers want partners who can run structured tests, share results, and help build the case that their network delivers real incrementality. If you walk into a network planning conversation with a clear measurement framework, you have leverage that a brand running opaque attribution does not.
What to watch next
Three developments are worth tracking closely over the next two quarters.
First, the convergence of loyalty data and retail media targeting is accelerating. As covered in prior analysis of loyalty programme evolution, retailers are moving toward AI-driven personalisation at the SKU level. When that targeting capability feeds directly into retail media placements, the relevance of those placements improves and the incrementality case gets stronger. Brands with joint business plans built around the retailer's loyalty data will see better media performance than brands running parallel to it.
Second, in-store digital media formats are receiving significant capital investment from major grocers. The measurement standards for these formats are still being written, and the brands that participate in early pilots will have a structural advantage when those standards crystallise.
Third, the gap between the 39 percent of ad spend currently in retail media and the expected 50 percent represents a large pool of budget that will move in the next 18 to 24 months. Where it comes from matters: if it comes from brand-building budgets rather than trade spend, the implications for long-term brand equity are material and deserve a separate planning conversation inside your organisation before the migration happens by default.