Retail Media’s Most Important Metric Has a Methodology Problem

A study published this week found that iROAS, the metric retail media networks most commonly use to demonstrate advertising value, can vary by 6.5 times on the same campaigns depending entirely on how it is calculated. Eighty-three percent of the campaigns analyzed could shift from positive to negative without any change in actual performance. The research comes from Albertsons Media Collective, working with Ovative Group and professors from Northwestern University’s Kellogg School of Management.

What the Research Found

The study, titled iROAS Demystified, examined 42 retail media campaigns and identified specific methodological forks that drive the variance. The Kellogg researchers on the project were Brett Gordon, Charles H. Kellstadt Professor of Marketing, and Eric Anderson, Polk Bros. Chair and Professor of Marketing. Derek Nelson, senior director of retail media consulting at Ovative Group, told The Drum: “Customers doing nothing different, media doing nothing different, just putting everything together differently, and you end up with wildly different results.”

The paper identified two specific divergences that illustrate the problem. Whether or not historical brand sales were included as a matching feature could swing the average iROAS from $1.23 to negative $0.14, according to The Drum’s reporting on the study. The two approaches to calculating incremental revenue, observed sales performance versus a Bayesian structural time series model, diverged by an average of 90%.

Liz Roche, VP of media and measurement at Albertsons Media Collective, framed the finding as a governance issue rather than a technical one. “That level of variance is a strategic signal,” she wrote in a blog post accompanying the research. “When methodology can materially shift performance narratives, leadership and decision-makers must clearly and comprehensively understand what sits beneath the number.” The paper is not arguing that iROAS is broken. It is arguing that the same label covers meaningfully different calculations across networks, and that most brands are not asking the questions needed to interpret results with confidence.

Jordan Witmer, managing director of retail media at agency Salt XC, told The Drum the dynamic plays out across his clients regularly. Results still reflect how audiences are selected and campaigns are delivered, especially when systems are designed to show ads to shoppers already likely to buy. That pattern, where measurement captures correlation with purchase intent rather than genuine incremental lift, is precisely what the research is trying to surface.

Why This Lands Now

The timing reflects a specific pressure point. eMarketer projects U.S. retail media ad spending will reach $69.33 billion in 2026, up from $58.79 billion in 2025. Brands are allocating those budgets across a fragmented landscape where, according to the IAB citing eMarketer data, Amazon and Walmart combined command 84% of retail media ad investment, leaving the remaining 16% distributed across more than 160 U.S. networks, each operating with its own measurement methodology.

The Skai and Stratably 2026 State of Retail Media report, based on a survey of 166 retail media advertisers, found that only 15% of brands report strong confidence in their measurement, and that 52% said better measurement would most accelerate their retail media investment. Brands are not asking for standardization across networks as the primary unlock. They are asking for proof that the numbers they are already receiving are reliable. The Albertsons study is a direct response to that ask, and it explains why the answer is more complicated than most networks have acknowledged.

What Albertsons Is Claiming in Response

Albertsons Media Collective is not simply surfacing an industry problem. It is positioning itself as having solved it, at least within its own network. Roche described the methodology Albertsons applies: advanced matching grounded in predictive purchase behavior to construct test and control groups, evaluation of match quality before reporting lift, and time-series modeling where business context requires it. Randomized testing is available on request.

The network is also extending measurement to in-store media. At CES in January, Albertsons presented results from a beta campaign for Mondelez, promoting Sargento Cheese Bakes across 116 stores with integrated in-store digital screens and onsite and offsite media. The campaign delivered a $2.41 matched-market iROAS and a 14% lift in in-store sales, according to Albertsons. The company plans to expand its in-store digital display network by approximately 800 additional stores in 2026.

Whether advertisers accept Albertsons’ self-assessment of its own methodology is a separate question from whether the research itself is sound. Nelson told The Drum the paper represents “a push for transparency rather than a push for standardization.” Brands do not need every retailer to measure the same way. They need to understand what they are getting from each one, and that requires asking questions most are not currently asking.

What Brand and Retail Media Teams Should Take From This

The study’s implications differ by organizational role. For VP-level retail media buyers and heads of commerce media at CPG companies, the finding reframes the due diligence required when receiving iROAS results from any network. The metric’s face value is not the issue. The implication for retail media buyers is that understanding the methodology producing it is now a prerequisite for using the number to allocate budget responsibly.

For category managers and trade marketing teams who rely on retail media performance data to evaluate retailer partnerships and set investment priorities, the variance the study documents means that comparing iROAS across networks without accounting for methodology differences may produce misleading rankings. A network reporting a higher iROAS may simply be applying a more favorable methodology rather than delivering superior incremental performance.

Roche’s framing of the responsibility is worth noting. “Not every one of those brands has a powerhouse data science team who can really make sense of all this,” she told The Drum. The Collective is presenting itself as a steward of measurement clarity, a positioning that becomes more credible as the broader market wrestles with the trust deficit the Skai and Stratably research documents. For brands working across multiple networks, the more immediate question is whether they can name the methodology behind each iROAS number they are using to justify their next budget decision.

Conversations On Retail

Conversations On Retail is a gathering place and resource center for retail and CPG executives, built to make it easier to stay current, discover the technologies and solutions shaping the industry, and connect with the people driving it forward.

We publish news, views, and reviews from staff editors, contributing experts, and trusted partners. Some articles are developed internally, while others are submitted by industry contributors or adapted from interviews and recorded conversations with industry leaders.