Incrementality Theater: Why Retail Media Keeps Grading its Own Homework

Everyone in the industry says "incrementality," and it’s started to mean whatever the person using it wants it to mean. All the major retail media networks sell “incrementality measurement,” and every independent measurement company sells an incrementality product. On the brand marketer side, if you present a retail media strategy without saying the word, you look behind the times.

It’s incrementality theater.

Here’s why: incrementality measurement sold by the media seller is last-click attribution in new packaging. Better options exist, but the industry has been slow to adopt them. Most brand-side marketers have been trained to evaluate retail media through the dashboards the RMNs provide. However, this cobbled measurement solution hasn’t been designed to show brands what's actually working across their full media mix. It shows each retailer's performance inside its own walls.

Five Retailers, Five Dashboards, One Shopper

We all know last-click attribution is broken, but the emerging problem we’re discussing less is what's replaced it: siloed retailer views of performance.

Each RMN reports on what happened inside its own environment. Those reports rarely show how that retailer's media contributed within the context of the brand's full marketing mix, or even the full retail-only mix. What brands get instead is a patchwork of dashboards, each claiming success on its own terms, with no consistent way to see overlap, duplication, halo effects, or true incremental contribution across channels.

That creates a major decision-making problem. A brand may see strong results from multiple retailers at once, while also seeing strong results in paid social, search, and connected TV (CTV). But without a cross-channel measurement approach, it becomes difficult to answer the most important questions. Which investments are driving net-new demand? Which are capturing conversions that would have happened anyway? Where are partners reinforcing each other, and where are they simply taking credit in isolation?

This is where siloed measurement becomes just as limiting as last-click attribution. One narrows credit to the final touchpoint. The other narrows visibility to the final platform. Neither gives marketers the full picture.

If You're Going to Say Incrementality, Mean it

Three things separate incrementality measurement that earns its name from incrementality measurement that is theatre.

• Independence from the media seller: A lift study run by the platform selling the media is subject to the same conflict of interest as the attribution it was meant to replace. Real incrementality measurement requires a party with no stake in the outcome it's reporting.
• Cross-channel by design: Retail media doesn't operate in isolation, and measuring it in isolation is what created the siloed-dashboard problem in the first place. Lift has to be assessed across the full media system, including search, social, CTV, onsite retail media, and offsite activations, using a common business lens. Otherwise brands are just trading one walled garden for another.
• Experimental and modeled in combination: Geo holdouts and audience tests produce the cleanest causal reads but can't cover every channel and every campaign. Modeled measurement fills the gaps and captures effects that experiments miss. Each approach has blind spots the other corrects. Neither alone is enough.

When those three conditions are met, retail media can be measured alongside every other channel using a common standard: contribution to incremental sales, revenue or profit. That's the measurement executives are actually asking for when they ask whether the media is working.

Start With the Right Question

Brands optimizing against dashboards that overstate performance keep investing in the tactics those dashboards reward, whether or not the business is actually growing. That's measurement compounding in the wrong direction, and it keeps happening because bad measurement is still confident measurement. The numbers look precise and the reports arrive on time, but nothing in the output tells you the measurement itself is flawed.

Retail media networks will market their product. That's their job. The job on the brand side is telling the difference between a number that's measurement and a number that's marketing. One question cuts through: Who is measuring this, and what do they have to gain from the answer? Ask the question before you form your conclusion about what drove your results.

Vicky Harizanova is senior director, analytics at Ars X Machina (AXM), an independent media planning and buying agency.