Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Digital Marketing › Channel Overview

Retail Media Explained

What retail media is, how brands buy ads on retailers' sites and apps, what the retailer's data adds, measurement caveats and where it fits in a channel mix.
Guideintermediate11 min readUpdated

What retail media is

Retail media is advertising that a retailer or marketplace sells on its own properties, its website and apps, usually to the brands whose products it stocks. Some retailers also let advertisers use audiences built from their shopper data on other sites and platforms. Collectively, the advertising businesses retailers have built are often called retail media networks. Being stocked comes first: the decision to sell through retailers at all, and what that route gives up in margin and customer data, is covered in the wholesale versus direct-to-consumer comparison.

The channel exists because of where purchases happen. The FMCG marketing guide explains that shoppers choosing between brands on a retailer's search results page are influenced by what they remember, what is visible and what is on promotion, and that when people search for everyday products it is usually on a retailer's site, which is why retail media and product listings matter more than web search for those brands. The FMCG guide treats marketplaces and retail media as a primary channel for consumer goods, with a caution on its reports that this guide expands.

For a business that lists its own products on a marketplace, retail media is the advertising that sits on top of the listing, and the marketplace channel page for online sales covers it from that seller's side: sponsored placements bid against break-even after the marketplace's fees. The comparison of marketplaces and your own store sets out the wider trade of selling through someone else's shop.

Where the ads appear

Retail media placements vary by retailer, and each retailer documents its own. In general terms they fall into four groups.

Placement, in general termsWhere it appearsWhat it doesClosest equivalent elsewhere
Sponsored listings in search resultsThe retailer's own search results, mixed with or above organic resultsPuts a product in front of shoppers searching for the category or a rivalPaid search
Product and category page placementsBeside or within product pages and category listingsCatches shoppers comparing productsContextual display
Display and homepage placementsBanners and featured slots on the site and appReaches shoppers browsing, often for launches and promotionsDisplay advertising
Off-site audiencesOther websites and platforms, targeted using the retailer's shopper dataReaches people the retailer knows buy the category, away from its siteProgrammatic display and video

The first two groups sit at the point of choice, where the mental and physical availability framework places physical availability: ranking and visibility in search and retailer results are among its measures. The last two behave more like ordinary display and video and are judged in the same way.

How it is bought

Retail media is bought through the same routes as other media, and the media buying hub lists them. Self-serve buying inside the retailer's own interface is the route the hub names for retail media, with targeting, bids, budgets and creative under the advertiser's control and the trade-offs of platform-reported measurement and auction pricing. Larger buys are often negotiated directly with the retailer's sales team as packages. Off-site audiences may be available through a demand-side platform (DSP), where the retailer makes them available that way.

RouteHow it worksWhat the brand controlsWhat to watch
Self-serveThe brand or its agency buys in the retailer's own interface, usually in an auctionKeywords, products, bids, budgets, creativeReporting comes from the seller; auction prices rise with competition
Managed or negotiated packageThe retailer's team sells and runs a package of placementsThe brief and the package termsBundled placements of uneven value; reporting set by the seller
Through a demand-side platformThe retailer's audiences are bought programmatically alongside other inventoryTargeting, pacing and frequency within the platformData and platform fees; what the retailer releases for measurement

The media buying models and pricing guide sets out the pricing models. Sponsored listings are usually bought per click in the retailer's auction, as the marketplace page notes for marketplaces; display placements and off-site audiences are often priced like other display, per thousand impressions; and the pricing guide lists revenue share as a model found in some retail media. For brands sold through grocers and other large retailers, the FMCG guide adds that much of the budget is negotiated with retailers as trade spend, promotions and paid placements, so a retail media proposal often arrives as part of a wider commercial conversation rather than as a media plan.

What the retailer's data adds

The retailer knows what its shoppers search for, view and buy. That gives retail media two things most channels lack.

Sales-based reporting. The FMCG guide notes that retail media reports are useful because they come from the retailer's own sales data, which a brand selling through retailers otherwise sees late, in aggregate or not at all. A brand can see which campaigns were followed by purchases of its products at that retailer without building any tracking of its own.

Audiences built from shopping behaviour. A retailer can offer targeting based on what people have bought or browsed, such as category buyers or lapsed buyers of the brand, and each retailer documents what it offers. Audiences like these are tied to the category more directly than demographic or interest targeting.

Both come with a limit. The data belongs to the retailer. From the brand's side it is someone else's data, used under the retailer's rules and reported in aggregate, not First-Party Data the brand owns. The comparison of first-party and third-party data explains the trade between data a business collects itself and data it accesses through others, and a brand that relies on retail media for its customer knowledge learns about its buyers only as far as each retailer chooses to report.

Measurement caveats

The FMCG guide puts the main caveat in one line: retail media reports are the seller's view of its own ad product, they cover only that retailer, and they usually credit sales that would have happened anyway. Each part has consequences.

  • The seller marks its own homework. The retailer decides what counts as an attributed sale, the window in which it is counted and whether sales of other products from the brand are included. Ask for the definitions before reading any figure.
  • Reported return is not caused return. A shopper who searched for the brand by name and clicked a sponsored listing would often have bought the same product from the organic listing below it. The reported ROAS includes those sales. The comparison of iROAS and ROAS explains why only the return measured against a holdout shows whether a budget line is worth funding.
  • One retailer at a time. Each retailer reports on its own shoppers. The cross-platform reach and frequency guide groups retail media networks with the walled gardens: logged-in identity inside, no user-level data released outside, and reports that cannot be deduplicated against anyone else's.
  • Paid can replace organic. Sponsored listings on terms where the product already ranks well may shift sales from organic to paid without adding any. The marketplace page's advice applies: watch whether total sales hold as sponsored spend tapers, because a listing whose sales fall away is being bought rather than earned.

The remedies are the ones used for any channel. An incrementality test withholds the advertising from a comparable group, a set of regions or an audience, and measures the sales that disappear. A marketing mix model estimates retail media's contribution alongside every other channel, and the tutorial on preparing data for a marketing mix model covers assembling spend for every channel, retail media included, on the same weeks. The ROAS calculator is useful for the arithmetic, but run it on incremental revenue and margin, not on the retailer's attributed revenue alone.

Where retail media fits in a channel mix

Retail media is strongest where the purchase happens on the retailer's site and the brand's job is to be visible and chosen there.

SituationFitWhy
A brand sold mainly through retailers, where shoppers search the retailer's siteStrongThe ads sit where the choice is made
A new product with no reviews or organic position on the retailer's siteStrong, for a periodSponsored placement gives it visibility while it earns organic rank
A rival is taking sponsored positions on the brand's category searchesOften worth testingHolding Share of Voice on the retailer's results protects the shelf
Building memory among people not currently shopping the categoryPartialOff-site audiences can add reach, but on-site placements reach people already shopping
The product is out of stock or off promotion at that retailerWeakThe FMCG guide notes that media running while a product is unavailable wastes much of its effect
A brand not sold by the retailerUsually weakThe placements and reports are built around the products the retailer sells

Retail media does not replace the work of being remembered. The availability framework separates mental availability, built mostly by broad reach and consistent branding, from physical availability, which includes visibility at the point of purchase. Retail media is mainly a physical-availability tool with some reach at the edges, so a plan that moves its whole brand budget into it may find shoppers who see the brand on the shelf but do not think of it before they arrive. A brand that also sells from its own store faces a second choice, between ads on a retailer's site and Shopping ads on its own feed, which the comparison of retail media and Shopping ads weighs by where the click lands and who keeps the customer.

Evaluating a retail media proposal

A proposal from a retailer is a sales document from the seller of the product being evaluated. Ask the same questions of every one.

QuestionWhy it matters
What exactly is included: placements, audiences, off-site reach, and for how long?Packages bundle placements of uneven value
How is an attributed sale defined, over what window, and does it include other products from the brand?Definitions decide the size of the reported return
Can the brand run a holdout, by region, store group or audience?Without one, the incremental effect cannot be seen
What data will the brand receive, at what level of detail and how often?Data that arrives late or in aggregate limits optimisation and the mix model
How does the price compare with the brand's other routes to the same shoppers?Retail media competes for budget with search, social and trade spend
Is the budget coming from marketing, from trade spend or from both?Spend split across budgets is easy to count twice or not at all
Will the placements shift sales from organic listings rather than add new ones?Paid placements on terms the brand already wins add little
Does the expected incremental return clear the margin the product earns?A reported return above break-even can still lose money once sales that would have happened are removed

A snack brand weighs a retailer's offer (illustrative)

Illustrative example — not a real result

Brambleshot Snacks tests a retailer's package with a holdout

Brambleshot Snacks, an invented snack brand, sells through three grocers. The largest offers a retail media package for £120,000 over three months: sponsored listings on snack searches, homepage placements and off-site audiences of category buyers. Its proposal projects a return of £5 in attributed sales for every £1 spent.

Brambleshot asks how a sale is counted and learns that the retailer credits any purchase of any Brambleshot product within 14 days of an ad view or click. It asks for a regional holdout and the retailer agrees to a one-month pilot of £30,000, with placements shown in eight regions and withheld in four similar ones.

Measure from the pilotFigure
Spend£30,000
Retailer-reported attributed sales£150,000 (return of 5.0)
Sales growth in test regions against holdout regions4.1%
Extra sales implied by the holdout£54,000 (incremental return of 1.8)
Brambleshot's margin on retail sales38%
Margin on the extra sales£20,520

The holdout shows that about a third of the reported sales were extra, and the margin on them does not cover the spend. Breaking the results down by product, most of the extra sales in the test regions come from the two new flavours, while sales of the established range barely moved against the holdout. The pilot cannot go further than that. It switched every placement on and off together, so it cannot say whether the sponsored listings, the homepage placements or the off-site audiences did the work, and the retailer's split of attributed sales by placement is the same attributed measure the holdout just discounted. Brambleshot therefore treats the smaller package as a hypothesis to test, not a finding: £45,000 over three months, limited to sponsored listings for new product launches on category searches, with a request to the retailer for separate test cells by placement so that the next pilot can show which placement earns its spend. It also adds retail media spend to the data it is preparing for its mix model.

Before entering a pilot's figures, the tutorial on using the ROAS calculator for retail media sets revenue to the holdout lift and margin to the brand's share of retailer sales.

Common mistakes

  • Judging retail media by its own reported return. The seller defines the sale and the window; only a holdout shows what the ads caused.
  • Bidding on your own brand name where you already rank. Sponsored listings there often buy sales the organic listing would have made.
  • Treating it as the whole brand budget. On-site placements reach people already shopping; memory is built before they arrive.
  • Running ads when the product is not available. Out-of-stock or unpromoted products waste the spend.
  • Adding up reach across retailers. Each network reports on its own shoppers and cannot be deduplicated against the others.
  • Losing track of which budget pays. Retail media negotiated in trade terms and bought from marketing budgets is easy to double-count or miss.
  • Signing a package before asking for a pilot. A short test with a holdout costs less than a quarter of spend that cannot be judged.
  • Leaving retail media out of the mix model. A channel missing from the data has its effect credited to something else.

Frequently asked questions

Is retail media the same as advertising on a marketplace?

Advertising on a marketplace is one form of retail media. The term covers any retailer or marketplace selling advertising on its own sites and apps, from sponsored listings in its search results to display placements and audiences used on other sites. The marketplace guide on this site covers sponsored placements for sellers judged on contribution after fees; this guide covers the channel more broadly, including brands that sell through retailers rather than listing themselves.

Can a brand that does not sell through a retailer use its retail media?

It depends on the retailer. Retail media is built around the products a retailer sells: the placements sit beside them and the reports count sales of them, so it suits brands the retailer stocks. Whether a particular retailer sells placements or audiences to other advertisers, and what its reports could show them, is something to check with that retailer before planning on it.

Should we trust the return a retailer reports for its own ads?

Treat it as one input. The report comes from real sales data, which is valuable, but it is the seller describing its own product, it covers only that retailer and it usually credits some sales that would have happened anyway. Ask how it counts a sale, over what window and whether it can support a holdout, then compare the reported return with what a test shows and with your mix model before scaling the spend.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Related