A buyer running offsite retail media this quarter can prove, to an audited standard, that the impression was viewable. They can prove it drove an incremental sale, using one of three methods the industry has written down. They can have the retailer’s measurement methodology disclosed and filter out invalid traffic to MRC rules.
Ask which company sold that impression, how many parties were paid along the way, or whether a second network sold them the same one, and the standard goes quiet.
That gap has a shape. Retail media built its measurement standards first, thoroughly and well. It has not yet built a supply standard, and offsite, where the retailer’s shopper data rides on programmatic inventory the retailer does not own, is where one is needed most. Until it exists, curation is how a buyer closes the gap themselves.
The data is the retailer’s. The inventory is not
Offsite retail media is a simple idea with a complicated supply chain. A retailer takes what it knows about its shoppers and uses it to target ads on inventory it does not own: open web, connected TV, social. The audience is the retailer’s. The impression is somebody else’s.
The money behind it is no longer a rounding error. Commerce media took $63.4 billion in US ad revenue in 2025, up 18% on the year and 21.5% of a record $294.6 billion digital market, on the IAB and PwC’s count. Onsite placements are finite and increasingly expensive, so the growth is moving off the retailer’s own pages and into everyone else’s.
That changes what a buyer is paying for. On a retailer’s site the premium buys the shelf: the shopper, the moment, the property. Offsite, the premium buys the data, and the media underneath it is bought at market through the same exchanges, the same sellers, and the same resellers as any other programmatic impression. The shopper segment is the retailer’s. The path it travels is not, and nobody in the retail media stack has been asked to account for it.
That is not an accusation. It is a sequencing story, and the sequence is worth spelling out.
Source: IAB / PwC Internet Ad Revenue Report, FY 2025
The category standardized measurement first
Retail media is one of the better-governed young channels in advertising. The IAB/MRC Retail Media Measurement Guidelines arrived in January 2024. IAB Europe’s Commerce Media Measurement Standards, now at version 2.1, are backed by a certification program that has been V2-only since July. Separate incrementality guidelines and in-store standards sit alongside them.
Between them, they settle a great deal. Viewability at 50% of pixels for one second on display and two on video. General invalid traffic filtered as a requirement, sophisticated invalid traffic as a recommendation, with a disclaimer where it is not. A 30-day lookback as the default, flexible windows on request. Three accepted methods for proving incrementality. Disclosure of methodology, measurement partners, and data sources, with the offsite and onsite numbers reported apart.
What none of them settles is where an offsite impression came from. There is no line on sellers.json or app-ads.txt, nothing on the SupplyChain object, nothing on path length, nothing on direct against resold. The guidelines do use the phrase traffic sources, and it means site banner, onsite search, offsite social: the format, not the seller.
The clearest picture is the IAB’s own list of seven questions brands should put to a retail media partner. Viewability method, MRC accreditation, invalid traffic filtration, online against offline sales, attribution modeling, in-store proximity, privacy. Every one of them is about counting. Not one asks who sold the impression.
This is what happens when a channel grows up answering the question advertisers asked first. They asked whether it worked. The standards bodies answered, and answered well. Whether the path behind the impression was clean was a question for later, and later is now.
Sources: IAB/MRC Retail Media Measurement Guidelines, January 2024. IAB Europe Commerce (incl. Retail) Media Measurement Standards V2.1, May 2026.
Curated or arbitraged, at a premium
Two kinds of supply can carry the same retailer’s audience, and they will not look different in a report.
Curated supply earned its admission. It met explicit criteria before it entered the deal; the path to it is direct, the impressions are de-duplicated; and every party in the chain is disclosed. Arbitraged supply was repackaged and resold. The path adds cost and subtracts nothing you can name.
In offsite retail media, the difference costs more than usual, because the premium is attached to the data and the data does not care which path it travels. A shopper segment does not clean a supply chain. Put the same segment on an impression that reached you through one declared seller and on an impression that reached you through four, and the reporting will show one audience, one viewability rate, and one sales lift. The buyer paid a retail media price for both. Only one of them was worth it.
Duplication is the sharper version of the same problem. A shopper who is in two retailers’ audiences is in two retail media networks’ offsite campaigns, on the same open web pages, often through the same exchanges. The buyer may have paid two premiums for one person and one screen, and nothing in the reporting stack will say so.
Provenance in this category looks like what it looks like everywhere else: a declared seller, a path you can trace back to the publisher, an impression that appears once. Curation is a standard, not a segment, and in retail media it is currently the only place that standard is being applied.
A shopper segment does not clean a supply chain. Share on XThe gap is not a secret. It is on the page
The body writing the rules has already said as much. Page five of IAB Europe’s V2.1 standard notes that the lack of interoperability between commerce media networks “makes it difficult for media buyers to deduplicate investments,” and that in the long run advertisers will need to bring sales into their own attribution models and run the deduplication themselves.
Read that as a buyer. The standard that decides whether a network is certified has told you, in its own text, that nobody is currently checking whether two networks sold you the same impression, and that the job will eventually land on your desk.
That candor is to IAB Europe’s credit. It is also the clearest possible description of the space between a measurement standard and a supply standard. The first tells you the impression was real and the sale was incremental. The second would tell you it was sold once, by someone you can name, along a path you can trace. Retail media has the first, in a certified, audited form. It has written down, in the same document, that it does not yet have the second.
Three questions to ask before the next offsite plan
None of these needs a new standard to be answered. They need to be asked. We set out the 4 things worth evaluating in any supply partner in June, and these are what that framework looks like pointed at a retail media network.
- Which sellers are authorized to sell this retailer’s offsite inventory, and where is that written down? If the answer is a measurement partner rather than a seller list, the question has not been answered.
- The premium buys the data. What is it buying in the path? Ask for the hop count between the publisher and your seat, and whether any of those hops is a reseller of another network’s supply.
- If the same shopper reaches me through two networks on the same page, how would I know? The standard says you would not. Ask what the network does about it, and whether the answer is a mechanism or a hope.
Retail media knows the shopper better than any channel in the plan. A channel that close to the person deserves paths you can stand behind.