How to Measure Retail Media Incrementality
Most retail media campaigns aren't reported well. That's not about the campaigns themselves. It's a measurement problem, and one the industry has been slow to fix.
The standard metric, ROAS, can overstate performance and understate it at the same time. It overstates when it claims credit for sales that would have happened anyway: a shopper who was already searching for your product on Amazon, who would have bought regardless of whether a sponsored listing appeared. And it understates when it misses the sales your campaigns actually caused but couldn't track: the shopper who saw your ad online and bought in-store, or the lift on a different retailer driven by spend on another. Cross-retailer effects are real and well-documented, and platform-reported ROAS doesn't capture any of them.
The result is a measurement picture that's simultaneously too generous and too incomplete to make good decisions from.
Why retail media makes this harder
The structural problems compound this. Each retail media network runs its own attribution, with its own methodology and its own look-back windows. A campaign on Amazon is measured differently than one on Walmart Connect or Target's Roundel, which makes cross-retailer comparison effectively meaningless if you're working from platform-reported numbers. A 2024 study by the Path to Purchase Institute and Skai found that 70% of advertisers struggle to measure the incremental performance of their retail media. And that's before accounting for the fact that each platform has a financial interest in the results it reports.
Then there are the variables ROAS doesn't account for at all: promotions running simultaneously on the same product, inventory changes, competitors’ pricing, search rank changes. Any of these can drive a spike in sales that gets credited to a sponsored product campaign that happened to be running at the same time.
What incrementality measurement actually requires
Measuring true incrementality means answering a counterfactual: what would sales have looked like without this ad? That requires controlling for the variables that affect sales independently of media (seasonality, promotions, competitive pricing, inventory levels, search rank, etc.).
This is what separates causal measurement from attribution. Attribution works backwards from a conversion and assigns credit. Causal measurement isolates what the ad actually caused, after stripping out everything that would have happened regardless.
In retail media specifically, the measurement approach has to be built around the commerce environment, not just the media. A model that doesn't account for the fact that your product went on promotion the same week as your sponsored brands campaign, or that a competitor went out of stock and temporarily boosted your organic sales, isn't measuring your ads. It's measuring noise.
What incrementality measurement gives you
When you measure correctly, two things change.
The first is which campaigns are actually working. Research from Incremental's own data shows 43% of retail media spend delivers an iROI of 6x or greater, on par with the best-performing media channels. But that sits alongside campaigns that look strong on ROAS and deliver little or no incremental sales. ROAS alone can't tell you which is which. iROI, sometimes referred to as iROAS (incremental Return on Ad Spend), calculated as incremental sales divided by advertising spend, can.
The second is comparability across retailers. When every retailer's performance is measured through the same causal lens rather than their own proprietary attribution, you can make informed allocation decisions. Is this dollar going further on Amazon or Walmart? Platform-reported ROAS can't answer that. Standardized incremental measurement can.
What this means for how you measure
The measurement approach needs to operate at the speed and granularity of retail media decisioning. Daily signals, not quarterly reports. It needs to handle the commerce-specific variables that drive sales in ways legacy marketing measurement tools were never built for: SKU-level promotions, real-time pricing, inventory, competitive presence.
Seventy percent of retail media advertisers still struggle to measure incrementality accurately. The ones who solve that problem don't just get better reporting, they get a real basis for knowing where to scale and where to pull back, and that's the difference between optimizing for efficiency and optimizing for growth.
