Why retail media measurement still ignores 70% of UK retail

In-store sales measurement is one of the unsolved problems in retail media, and it is the part of an effective retail media strategy that most measurement frameworks quietly skip past.

According to the ONS, around 70% of the UK's retail transactions still happen in-store, rather than online. Across EMEA the picture is similar: physical retail still accounts for the majority of transactions in every major market. The retail media industry has spent the last five years building measurement frameworks for the other 30%.

That is the problem in one sentence.

We have written up what an effective retail media strategy looks like in full, with in-store measurement as one of the foundations. You can read the guide here. What follows is the case for why this part of the strategy matters, and what the measurement gap actually costs.

Where does retail media measurement go wrong?

Retail media networks have become the fastest-growing channel in digital advertising, and the dashboards have kept pace. Impressions, clicks, onsite conversions, ROAS within the retailer's own ecosystem, all of it measurable, all of it reportable, all of it neat. The trouble is that for most UK and EMEA retailers, the dashboard is describing a minority of the business. The majority of transactions triggered by a retail media campaign happen somewhere the campaign cannot see: in an aisle, at a till, on a different device, several days later.

If your retail media measurement stops at the click, you are not measuring retail media. You are measuring a slice of it and calling it the whole.

Why does retail media measurement struggle with in-store sales?

Closing the loop is the industry's stated ambition: connect the ad exposure to the eventual sale, regardless of where that sale happens. In practice, "closed-loop" usually means closed within a single channel. Display impression to website visit. CTV view to app open. Onsite ad to onsite basket.

Find out more about Epsilon Retail Media’s onsite enhancements through the acquisition of CitrusAd.

The genuinely closed loop, the one that joins a display impression on a Tuesday to an in-store transaction on a Saturday, requires something most measurement stacks do not have: an identity layer that resolves the same person across devices, sessions, and the physical store. Offline attribution depends on this. Cookies cannot do it. Email-based identity solutions cannot do it reliably either, because one customer with four email addresses appears as four customers, and an in-store shopper who never gives their email at the till appears as nobody at all.

The result is a measurement gap that systematically under-reports the channel's actual contribution. Retail media budgets get justified on the visible 30%, and the invisible 70% gets credited to "brand," "halo," or nothing at all.

What does in-store measurement actually look like when it works?

When Currys ran a multi-channel campaign across retail media, CTV and online video with identity resolution underneath it, the measurement showed 56,000 incremental store visits, a 20% in-store sales increase, and a 46% conversion uplift compared to display alone. None of that would have surfaced in a click-attributed report. The campaign would have looked like a CTV spend with a soft ROAS, and the retail media team would have been arguing for budget on the wrong evidence.

The reason those numbers held up under scrutiny is straightforward. The same identity that resolved the digital impression also resolved the in-store transaction at the till. The person who saw the ad and the person who bought the dishwasher three days later were the same person, demonstrably, in the data. Not an inferred match. Not a cohort overlap. The same individual.

This is what changes when identity is the foundation of in-store sales measurement rather than the workaround. The performance does not change. The visibility of the performance does.

What does effective retail media measurement require?

The honest answer is that retail media measurement that captures the full 70% does not start with measurement. It starts with identity resolution.

Without a persistent identity layer that resolves to the individual, not the device and not the email address, every other measurement claim is built on sand. Loyalty and first-party data have to feed that identity with high-confidence signals, including in-store transactions, which most measurement frameworks still treat as out of scope. And the methodology that connects exposure to outcome across channels has to be verified independently. Epsilon's outcomes measurement is MRC accredited, which remains rare in the category and matters precisely because closed-loop claims are easy to make and hard to validate.

This is the strategy underneath the measurement problem. We have written it up in full, with the supporting framework and the pieces that fit alongside identity resolution to make retail media measurable end-to-end.

The question to take to your next planning meeting is not "what is our retail media ROAS." It is "what share of our retail media-influenced sales is our measurement actually capable of seeing." If the honest answer is 30%, the dashboard is not a measurement system. It is a sampling system, and the sample is the easy part.

What else does an effective retail media strategy need?

In-store sales measurement is one part of getting retail media right. The full strategy depends on two more capabilities that work alongside it.

The first is finding the right shoppers in the first place. Most retail media budgets reach the same crowded pool of high-intent buyers across every campaign, which inflates frequency without inflating sales. Reaching new and unique shoppers requires identity that can find people who are genuinely in-market, not just people who happen to be addressable.

The second is activating across channels without breaking the identity thread. Retail media is no longer onsite-only. It runs across CTV, display, online video and increasingly into the open web. The campaigns that work join those channels through the same identity layer, so frequency, sequencing and measurement hold together. The campaigns that do not work treat each channel as a separate buy and end up paying for the same shopper several times.

In the full guide we cover all three: finding unique shoppers, activating seamlessly across channels, and measuring real sales impact across online and in-store. With practical guidance on what to look for in a partner who can deliver against each.

Read the full guide, 3 elements of a successful retail media strategy