Retail media measurement: absence of evidence is NOT evidence of absence

Retail media conference headlines and panel sessions always end up talking about measurement in retail media and how it’s ‘broken’, or how ‘measurement is not comparable with other media channels.’ What is the reality behind these claims? Colin Lewis investigates.

In 1476 or 1477 , the printer William Caxton issued the first recognised advertisement in the English language. The ad offered for sale Caxton’s edition of the ‘Sarum Ordinal’ or ‘pye’ – the priest’s manual of variations of the bible during the ecclesiastical year. The advertisement was intended to be displayed in the neighbourhood outside Caxton’s shop in Westminster Abbey.

Advertising has come a long way from selling bibles ‘good chepe’, but I would guess that even Mr Caxton wondered exactly how many advertisements he needed to hang out in the neighbourhood to get a return on his investment.

Advertisers have always had the same problem: how to capture information on the outcome of an advertising message and whether there was a better way they could have spent their money?

Even if an advertiser could afford the $8 million to $10 million for a 30-second slot at Super Bowl 2026, there is still the question of return on investment.

Measurement matters, as it is the currency by which advertisers make choices about whether to buy a particular media.  And media measurement is a story that goes back not as far as the 15^th century, but still close to 100 years ago.

The origins of media measurement

Audience measurement, often called “ratings” in TV and radio, emerged in the 1930s. Broadcasters were selling airtime, and advertisers needed a credible, comparable way to estimate how many people were reached by a programme or time segment.

The first ratings measurement process was not a TV or radio station initiative: the ‘Cooperative Analysis of Broadcasting’ (CAB) was backed by advertisers and agencies to create a measure that enabled them to pick and choose among broadcast outlets and provide authoritative measures of audiences. The CAB worked with a man named Archibald Crossley to create this measure. He had decided to measure ‘exposure’ in his radio ratings analysis – who listens, for how long, and with what regularity.  The exposure measure that Crossley championed quickly became the standard for measuring ratings.

Crossley’s legacy is that he helped formalise a model that still underpins media “currency” today: exposure is treated as the measurable unit, estimates are produced from probability samples, and results are reduced into standardised metrics that can be traded consistently across outlets, programmes, and time periods.

Over time, this approach made it possible to price audiences at scale using shared rules, and to manage investment decisions with a common reference point. The measurement business turned into multi-billion-dollar business in countries all over the world, each with its own set of tools for measurement. 

However, even with this decades-old head start, media channels such as TV, radio and outdoor are still decidedly ‘old-school and beset by the problem of the gap between the actual and the measured audience. 

TV, radio and outdoor measurement

You might think that TV, radio and outdoor have much more robust measurements than retail media. You might be wrong – let’s see how they are currently measured.

• TV ratings measurement: In markets like the UK (Barb) and Australia (OzTAM), TV ratings measurement rely on a relatively small, carefully recruited panel. The TV Barb panel in the UK consists of 7,000 households, representing approximately 16,000 individuals. With panel-based TV ratings, the measurement is built on a small set of volunteer homes that have metering equipment installed to track what’s on the screen, and an extra step to label who in the household is actually “watching”. That “who’s watching” step is usually self-declared, so the system can record a viewer even when the person has wandered off, fallen asleep, or is staring at TikTok videos while the TV plays in the background.
• Radio measurement has historically relied on diaries, and the central issue is the same as TV: estimating listening behaviour from a sample rather than observing every listener. In the UK, the established approach is
  • a 15,000 quarterly sample completing a 7-day diary, with a choice of online, mobile or paper
  • a continuous panel of 5,000 participants providing app-based passive listening data
  • a targeted diary panel of around 1,000 to boost representation of young people and selected ethnic minority groups.
• Outdoor advertising measurement is modelled using calculations such as “likelihood to see” within defined visibility areas, using short time increments and applying additional new measures for dynamic content and illumination. Out-of-home measurement typically blends travel or movement data, independent count or traffic inputs, and visibility or attention adjustments to produce impressions intended to reflect probable viewing rather than mere opportunity.

I think you can agree that these measurement tools must work, as they are agreed on by both the media industry and the advertisers.

Three critiques of existing channel media measurement

There are three critiques that I can level at the measurement approaches used across these channels:

• Estimates based on samples: All of them are mathematical models built on top of a very small sample of actual human behaviour. That does not make them “wrong”, but they are estimates. The sample is so small (0.01% of the population) that any error at the household level is magnified.
• No sales attribution: None of these media channels offer closed-loop attribution, where the advertiser can directly connect the measurement provided by the channel to a sale.
• Exposure is not attention. These systems don’t fully account for the fact that the TV viewer or radio listener may not be giving full attention, because people multitask (for example, scrolling TikTok or Twitter) while the TV is on, or they watch later.

Each measurement system is built to standardise audience delivery for buying and selling media, not to prove the incremental sales impact of a specific spot or poster in isolation.

As I often say in my keynotes, brand advertisers have not historically gone to TV, radio, or outdoor media owners and asked them to prove how the ad they bought last night, last week, or last month contributed to sales, or to provide a definitive ROI measure.

The hypocrisy of retail media measurement criticism

Let’s start with quotes from recent reports and article on retail media:

• “The retail media boom is doomed if measurement isn’t fixed. Brands aren’t going to be able to keep increasing that amount of money into these retail media networks unless they come up with better measurement solutions.”
• “Retail media is still an untapped landscape with no real measurement parameters. The attribution is different, measurement capabilities are different, and if a retail media network isn’t full funnel, it’s harder to put them into an MMM.”
• It’s tough to understand how all your media is performing together with so many walled gardens. Our biggest challenge with retail media measurement is that there’s no interoperability across retail media networks.

A quick scan of these statements, combined with an examination of how other media channels are measured, leads to one conclusion: retail media is held to a higher standard than almost any other media channel.

This is striking given that other channels (for example, TV) still take a much higher share of spend in absolute terms. Demands for ROI and ROAS from most other channels, at the level retail media is expected to provide, largely don’t exist.

You might say that TV and other channels are higher impact and measurement capabilities like econometrics studies can be applied, so my criticism is not valid. 

Fair enough: but then I will point out that these measurements are NOT paid for or provided by these channels. They are the responsibility of the brand advertisers. ‘Caveat Emptor’ must be the thinking by the channels.

No “apples-to-apples” comparisons

Ryan Verklin is the paid and retail media lead at Bayer in the US. He points out that for brands trying to balance national brand investment against performance-driven commerce channels there is a problem: “We’re not really having apples-to-apples comparisons. You can’t really compare it, sometimes, with national campaigns.”

The problem cited is that every network measure success differently. “Right now, all the commerce media networks have their own measurement silo. We need a holistic view of measurement instead of the silos that currently exist.”

This sort of statement is a real headscratcher. TV, radio and outdoor cannot be compared apples-to -apples and never have been.  Yes, one might argue that we are talking about retailers here, so they should all have a common ‘currency’.

The thing is THEY DO have a common currency: The IAB Europe has issued standards for both online and in-store retail media measurement over the past couple of years. The latest publication (Commerce Media Measurement standards V2) was published in January 2026 and covers both traditional retail as well as quick commerce platforms.

IAB Europe has also developed a certification programme. This programme has recently expanded and been launched in Australia.

In the US, the MRC in collaboration with IAB US has provided a standardised framework for consistent, transparent reporting on retail media, focusing on on-site, off-site, and in-store metrics. These standards emphasise closed-loop attribution, incrementality, and, as of late 2025, in-store, digital endcap, and quick commerce,

The standards have been put together by the industry, and many retailers are committed to them.  So, how can it be said that measurement does not exist?

Again, one might say ‘not all retailers adhere to these standards’.

Well, yes. But a brand advertiser can ask the retailer to adhere to these standards and to ask for a roadmap for when they will deliver on these standards.  I know of at least three major brand advertisers who tell their Retail Media Networks what measures they want and what this will unlock in terms of budget if they deliver on the measurement roadmap.

Nothing is in a vacuum

As Russ Deiringer of Stratably writes: “Retail media search does not operate in a vacuum: discounts, promotions, subscribe and save, seasonality, non-retail media investments, and other factors meaningfully impact retail media performance. This is very true of key selling periods like Prime Day, when a higher portion of sales are driven simply by the increased traffic of ready-to-buy shoppers compared to how much ads are estimated to have contributed based on LTA.”

Saying ‘retail media measurement is bad’ discounts everything that is happening in the market and in the wider world.

I wonder: are these brand advertisers mixing capabilities up? They see what Google, Meta, TikTok, Amazon can do in terms of measurement and metrics and wonder why retailers do not have a standard metrics that are comparable.

The reality is that Amazon, Google and Meta took years to have anything but the most basic metrics. And they are platforms that work in every market in the world, as opposed to be tailored to an individual country.

We also live in an imperfect world: Retailers are just that – retailers. They are not technologists; they do not have armies of highly-paid data scientists on call like Google or Meta do. Their technology vendors don’t provide them with the exact reports that every advertiser needs. They would struggle to pay the salaries that AI and data scientist can demand – as they have pay scales like any other business.

Brands – get your own house in order

The common theme among all the retailers I deal with is that each brand wants different things, each brand briefs differently, each brand measures success differently.

Every large advertiser already has its own MMM, incrementality standards, attribution logic, and definitions of success. This fragmentation is not a retail media problem. It exists across TV, digital, OOH, and sponsorship. Retail media is simply where it becomes visible because the data is closer to sales.

Retailers own the transaction data, so future is shared measurement frameworks where responsibilities are split, rather than complaining about the retailers.

The reality is that retail media measurement will not be stable until brands standardise their own internal success frameworks and retailers standardise their measurement infrastructure. One without the other fails.