Over the last decade, retail media at many brand organizations has evolved from a unique new advertising channel to a cornerstone of commerce marketing strategy. But while activation — and the investments supporting it — have increased dramatically, standards and practices for measuring the true business impact of retail media haven’t evolved at anywhere near the same pace.
Measurement is a vastly important topic of discussion for the industry as brand advertisers increasingly require quantifiable evidence of real business impact that will justify retail media investments within their marketing organizations — and subsequently use measurement capabilities as a decision-guiding point of comparison when evaluating network partners.
Leading these efforts at Publicis Commerce are VP of Analytics Prabhpreet Sidhu and SVP of Media Willy Blesener, who collaborate to develop strategies, practices, and processes that ensure all retail media activation executed on behalf of our clients leverages the best-in-class measurement techniques needed to understand true business impact.
The following article is the first in a series in which Sidhu and Blesener will examine some of the most critical issues surrounding retail media measurement. These questions represent the key issues that industry practitioners are facing as they work strenuously to close the gap between retail media activation and measurement.
Q: How should brands measure true incremental impact?
Blesener: This is an extremely important topic that remains unresolved due to the lack of alignment on even the definition of incrementality. If I’ve purchased the same exact product at Kroger for 20 years, and then I buy it for the first time on Amazon, I’m obviously not an incremental conversion across the totality of that brand’s business (although there are still useful things that can be done with this information).
At the moment, everyone — retailers, agencies, third-party data management platforms, tech vendors — is racing to develop the right solution to solve an ambiguously defined term with a lot of subjective considerations. But a leap of faith may be required for non-technical decision makers who don’t live and breathe data science to trust the complex outputs they’re receiving. Part of the long-term solution is education, but until brands become comfortable with the methodology — or data scientists start controlling ad budgets — we’ll need to continue asking them to take a leap of faith on hard-to-interpret data.
Sidhu: Generally speaking, incrementality refers to the additional impact in sales, conversion, or other shopper action generated by a specific marketing activation beyond what would have happened without it. The goal of incrementality is to measure true effectiveness. We should be aiming to prove causal lift, not just correlation.
These days, finance teams will only sign off on incremental ROAS or another incrementality measurement when the experiment is transparent and statistically sound. So A/B testing, geographic holdouts, and well-specified regression models are the go-to tools right now.
The definition and the methodology, therefore, are equally important. We’re still in the early stages of building standards and effective methods of measurement, but the core concept is to measure impact through causation and not just correlation.
Q: Are retail media networks getting better at providing the incrementality measurement that advertisers are seeking?
Sidhu: More and more retailers are offering incrementality as part of their reporting options and becoming more transparent about their methodologies. The challenge, of course, is that these methodologies may be inconsistent from one retailer to the next — A/B testing versus regression-based modeling, for instance. (How do you compare performance across retailers for incrementality? By standardizing the data.) We still have a long way to go, but we’re headed in the right direction.
Blesener: Yes, transparency has increased tenfold. The reality, however, is that the level of transparency often depends on performance: If it’s a viable metric whose output is likely to increase the advertiser’s future investment with the retailer, it will be released. But if that same metric proves to be a detriment to increased investment, brands aren’t likely to get it. We therefore should be cautious about what we receive and validate what we can’t. Brands shouldn’t treat these metrics as their sole source of truth.
Sidhu: Ultimately, brands should be doing their own analysis on incrementality instead of simply asking retailers to provide it. Ask for the raw data and do your own modeling so that the methodology is more consistent across retailers and you can make a true apples-to-apples comparison. That being said, a robust martech stack that includes connected identity like Epsilon’s CoreID to track the customer journey across platforms and retailers is essential for accurate measurement.
About the Authors
Prabhpreet Sidhu brings a technically diverse background and wealth of experience to his role as VP-Analytics for Mars United Commerce, where he excels in delivering impactful data analytics solutions. With over 10 years of solutions architecture experience at leading organizations such as The Walt Disney Company, he has developed groundbreaking business intelligence and artificial intelligence solutions, including real-time automated measurement, multi-touch attribution, and customer segmentation.
Willy Blesener joined Mars United in January 2025 to take charge of the company’s highly regarded and rapidly growing commerce media practice. His extensive background includes more than a decade dedicated to retail media, most recently with Omnicom and Flywheel helping to build the agency’s retail media analytics practice. Prior to Omnicom, Blesener worked across agencies within Advantage Unified Commerce, spearheading retail media activation and execution.
