What Is a Retail Media Network (And Why Every Retailer Suddenly Has One)

A retail media network is a retailer selling advertising against its own shopper data and its own digital shelf. That is the whole idea. Everything else, the acronyms, the self-serve platforms, the clean rooms, the “commerce media” rebrand, is implementation detail layered on top of that one sentence.

The reason the category exploded is less romantic than the conference talks suggest. Retail is a low-margin business. Grocery runs at 2 to 4 percent operating margin on a good year. Advertising runs at 60 to 80 percent gross margin. When a retailer discovers it can sell a sponsored listing on a page it already owns, to a supplier it already invoices, using data it already collects, it has found the highest-margin product line in the building. Amazon proved the model at scale. Everyone else spent the following years catching up.

What a retail media network actually sells

Strip away the packaging and there are four inventory types.

1. Sponsored product listings. Paid placement inside search and category results on the retailer’s own site or app. Sold on CPC, auction-priced, and usually the first thing launched because it is the closest thing retail has to a money printer.
2. Onsite display and banners. Homepage takeovers, category page banners, brand storefronts. Sold on CPM or as fixed-fee placements. Lower intent than search, better for brand-building budgets.
3. Offsite media. The retailer takes its first-party shopper segments and activates them on Meta, YouTube, connected TV, or programmatic display. The impression is served elsewhere; the targeting and the measurement come from the retailer.
4. In-store and retail-adjacent. Digital screens, shelf-edge displays, sampling, receipt offers, checkout media. The hardest to measure, the most physically constrained, and the piece most retailers keep promising for “next year”.

If you are trying to work out which of these to fund first, that is a real strategic decision with real trade-offs, and it deserves its own analysis. We wrote one: onsite versus offsite retail media.

Who is actually paying

This is where operators get confused, because the money often does not come from a media budget at all.

In consumer goods, spend historically sat in trade budgets: the money a brand pays a retailer for shelf position, promotional features, listing fees, and co-op advertising. Retail media is, in practice, a partial reclassification of trade money into media money. That has three consequences worth internalising:

• The buyer is frequently a category or key account manager, not a media planner. They think in terms of shelf, distribution, and volume, not reach and frequency.
• The negotiation happens inside the annual joint business plan. Retail media budget is often committed months before any campaign is briefed.
• Because it started as trade money, the measurement expectations are commercial: sales lift, share gain, distribution. Not viewability.

The mature networks eventually pull in genuine incremental media budget from brand teams, but that only happens once the measurement is credible. Which is the entire problem.

How the money flows

A simplified path for one euro of brand spend:

1. Brand commits budget through a joint business plan or a self-serve platform deposit.
2. Retailer serves the impression or click against its own inventory (onsite) or buys it externally (offsite).
3. If offsite, a media cost leaves the building: the retailer pays the platform or the exchange. Onsite has effectively zero incremental serving cost.
4. Retailer recognises revenue, reports performance back to the brand, usually with attributed sales pulled from its own transaction log.
5. Brand decides whether to renew.

The gross margin difference between step 3 onsite and step 3 offsite is why almost every network launches onsite first. Onsite is close to pure margin. Offsite is a media reseller business with a thinner spread, but it scales beyond the ceiling of your own traffic.

Why every retailer suddenly has one

Four forces converged.

Signal loss. Third-party cookies degraded, mobile identifiers got restricted, and logged-in first-party purchase data became the scarcest asset in advertising. Retailers happen to sit on exactly that.

Margin pressure. Cost inflation squeezed retail P&Ls at the same moment a high-margin revenue line appeared. Boards noticed.

Closed-loop measurement. A retailer can tie an ad impression to an actual basket, not to a modelled conversion. That is a genuinely better measurement story than most of digital advertising, provided you do not confuse attribution with causality (more on that below).

Vendor infrastructure. Launching a network no longer requires building an ad server. Licensed platforms mean a mid-sized retailer can be live in months.

The trap: attribution is not incrementality

Almost every retail media network’s default report is last-click attributed sales. It looks spectacular. Return on ad spend of 8x, 12x, sometimes higher.

Most of it is not real.

A sponsored listing on a branded search term will show a huge attributed return, because the shopper typed the brand name and was going to buy anyway. You have not created demand; you have taxed it. The measurement question that matters is what would have happened without the ad, and answering it requires holdouts, geo tests, or matched-market designs rather than a click log. If you take one operational discipline from this piece, make it that one, and read how to measure retail media incrementality before you sign a renewal off attributed ROAS alone.

A five-question readiness test for retailers

Before you launch anything, answer these honestly:

1. Traffic. Do you have enough logged-in or identifiable sessions to make onsite search auctions competitive? Thin auctions produce low CPCs and disappointed sellers.
2. Data. Can you join an ad exposure to a transaction at the individual or household level, within a defined window, reliably?
3. Commercial ownership. Who owns the P&L: the trade team, a new media unit, or a joint venture? Ambiguity here kills more networks than technology does.
4. Shopper experience. How many sponsored slots can a results page carry before conversion rate drops? Every network eventually finds this ceiling; find it deliberately rather than by accident.
5. Reporting credibility. Can you produce a measurement report that a brand’s finance team would accept without a fight?

If three or more answers are shaky, you do not have a retail media business yet. You have an inventory idea.

What good looks like at 18 months

A network that is working tends to show the same signals: onsite search sold out on the top categories, a defined and published rate card, a self-serve platform handling the long tail of smaller suppliers, at least one incrementality study per major category per year, and a repeat rate above 70 percent among top-20 suppliers. Revenue is nice; repeat rate is the truth.

Retail media is not a hype cycle that will pass. It is the logical consequence of retailers owning the two scarcest assets in modern advertising: purchase data and the moment of purchase. The open question is not whether to build one, but whether yours is priced, measured, and staffed like a real media business or like a side project bolted to the trade calendar.

The pan-European definitions behind all of this, the three environments and three buying models, and the organisational fault lines that decide who captures the growth, are in The Third Wave of Digital Marketing.