IAB says legacy measurement is cheating retail media out of its real value

The Interactive Advertising Bureau published a white paper on April 7, 2026, arguing that Marketing Mix Modeling - a measurement methodology that has guided advertising budget decisions for decades - is structurally ill-suited for the retail media era and is causing consumer packaged goods brands to undervalue one of their most productive channels.

The paper, authored by Collin Colburn, Vice President of Commerce and Retail Media at IAB, and Priyash Shahane, Manager of Measurement Science at Instacart, does not call for the end of MMM. The argument is narrower and more technical: MMM was designed for a different media environment and, when applied to retail media without adaptation, produces outputs that systematically misattribute sales and push brands toward suboptimal budget decisions.

Commerce media spending is growing at pace. According to the paper, IAB forecasts 12.1% growth in commerce media spend for 2026, a trajectory that makes measurement accuracy an increasingly consequential question. With more budget flowing into retail media networks, the gap between what models say and what actually happened widens - and the stakes attached to that gap rise.

Why MMM worked in the first place

MMM has been used since at least the 1960s. Its core mechanism is straightforward: by examining historical, aggregated data across channels, the model estimates how changes in spend correlate with changes in business outcomes over time. It was built for environments where direct measurement was impossible - traditional television, radio, out-of-home - channels that produce no clickstream data and whose effects on sales can only be inferred statistically.

The approach succeeds when there is observable variation in spend. A brand runs a TV campaign for four weeks, then pauses it. Sales go up, then flatten. The model can parse that signal. It also succeeds when sales data is concentrated in trackable locations, typically point-of-sale systems at a limited set of brick-and-mortar retailers. Under those conditions - reach-based channels, offline sales, spending that turns on and off - MMM delivered usable estimates despite its reliance on correlation rather than causation.

Privacy changes have reinforced its relevance in recent years. As addressable signals erode and deterministic tracking becomes harder, MMM offers a way to measure channel performance without relying on individual user data. That structural advantage has kept it at the center of many measurement stacks even as the media landscape shifted underneath it.

Three ways retail media breaks the model

The IAB paper identifies three distinct structural mismatches between how MMM works and how retail media operates.

The first involves spending patterns. Retail media is designed to be always on. Budgets rarely stop or vary in ways that classic MMM can detect. According to the paper, MMM requires spend variation to reveal impact - no variation means no measurable incrementality, and the model pushes that activity into the baseline. It simply reads constant activity as a background condition rather than a driver of outcomes. Buying dynamics compound the problem. Traditional media is planned on CPMs, impressions, or GRPs - inputs that translate cleanly into MMM structures. Retail media introduces CPC, CPA, and algorithmic auction dynamics that shift multiple times within a single day. These modalities do not map onto the exposure-based inputs MMM expects, meaning their performance is structurally underweighted. The consequence, according to the paper, is that retail media looks inefficient inside the model, and brands reduce investment in a channel that may actually be performing well.

The second mismatch involves data coverage. Most retail media networks sit on top of fragmented commerce ecosystems. The paper cites Instacart as an example: the platform connects more than 2,200 retail banners on the Instacart Marketplace with 310 or more grocery ecommerce sites. MMM, however, is typically fed only point-of-sale data from a limited set of physical retailers. When a retail media network drives national, omnichannel sales - delivery, pickup, marketplace transactions, cross-retailer spillover - MMM cannot parse the signal from the noise. The lift is real but has nowhere to land inside the model. The paper states that MMM does not attribute enough to retail media because the underlying sales impact is diversified across data sources the model does not ingest, or because those sources have too small an individual impact to achieve the signal sufficiency the model requires.

The third mismatch is perhaps the most fundamental: MMM was designed to estimate impact in environments where direct measurement did not exist. Retail media now provides the opposite. According to the paper, closed-loop, SKU-level, transaction-level measurement is available. Correlational modeling is not only unnecessary in this context - it is outdated. While MMM infers impact, closed-loop reporting and incremental sales lift testing prove it causally, accounting for other factors. MMM also typically aggregates all retail media into a single tactic, obscuring campaign-level performance differences that closed-loop systems can capture with precision.

The iROAS dimension

Central to the paper's argument is the concept of incremental return on ad spend - iROAS - which distinguishes between sales that occurred because of advertising and sales that would have happened regardless. This is the measurement question that closed-loop retail media systems are positioned to answer directly, and that MMM can only approximate through inference.

According to the paper, brands can end up making strategic investment decisions based on estimates when factual, causal evidence is already available from the retail media networks themselves. The paper quotes Kimberly Sugden, former Senior Marketing Manager at PepsiCo: "The future of commerce marketing is here - where advanced causal closed-loop retail media measurement reveals the true incremental sales impact of campaigns. By grounding strategic decisions in this level of data-driven accountability and iROAS clarity, brands can invest with confidence, optimize media with precision, and accelerate growth."

The IAB has done prior work in this area. The paper references the "Guidelines for Incremental Measurement in Commerce Media," a document released by IAB and IAB Europe on November 3, 2025, which established standardized frameworks for commerce media measurement and defined incrementality as "the causal impact of marketing by identifying the additional business outcomes directly driven by a campaign or tactic, compared to what would have occurred in the absence of marketing activity." That earlier document, covered by PPC Land at the time of publication, set much of the conceptual groundwork that the new April paper builds on.

What the paper recommends

The paper outlines three recommended shifts for CPG marketers, framed not as replacements for existing infrastructure but as adjustments to how different tools are prioritized.

First, it suggests repositioning MMM within the measurement stack. MMM retains value for long-range portfolio planning, budget guardrails, and high-level mix scenarios. What it should not do is dictate the precise level of investment allocated to closed-loop channels. The paper is explicit: MMM should complement primary platform-level measurement, not compete with it.

Second, it recommends adopting measurement built for closed-loop environments. Incrementality testing - control and exposed designs, randomized experiments, geographic lift tests - should serve as the primary foundation, not optional enhancements. These methodologies are purpose-built for environments where media and commerce are tightly integrated and provide the causal clarity that MMM is not designed to produce for retail media.

Third, it recommends auditing current decision logic. According to the paper, if MMM consistently under-credits retail or commerce media relative to platform-reported or experimentally validated outcomes, the issue is almost always a measurement mismatch rather than a media performance problem. Large gaps between MMM outputs and causal retail media results are a signal to expand model inputs, adjust specifications, or redefine how commerce signals are captured.

Colburn frames the challenge in methodological terms: "As media channels evolve from inferred exposure to observable outcomes, measurement has to evolve with that. The most effective marketers will no longer ask 'which methodology is best?', but 'which methodology is appropriate for the signal, channel, and decision at hand?'"

Context: a measurement debate with long roots

The IAB paper arrives at a moment when the broader measurement landscape is actively contested. Research published in February 2026 found that commerce media follows gaming as one of the most underrepresented channels in marketing mix models, with 50% of MMM users reporting it as underrepresented. That same research indicated marketing measurement confidence stalled in 2025, with 54.1% of marketers reporting no change year-over-year and 14.3% saying confidence declined.

The tools themselves are evolving. Google's open-source Meridian MMM, which became globally available in January 2025, received updates in September 2025 to incorporate non-media variables such as pricing and promotions, channel-level contribution priors, and enhanced binomial adstock decay functions designed to capture longer-term upper-funnel effects. Prescient AI announced in July 2025 what it described as the first marketing mix model built entirely from scratch since the 1960s, with a framework designed to identify multiple efficiency points rather than assuming linear ad saturation.

The question of who builds the model has also come under scrutiny. A PPC Land analysis of Meta's Robyn MMMexamined the structural dynamic that emerges when a platform creates the tool that guides budget allocation - a dynamic that applies equally to any measurement system built or operated by an entity with a commercial interest in the outcome.

Instacart's own measurement infrastructure has grown considerably. The platform received Media Rating Council accreditation in March 2024, with the MRC validating accuracy across Sponsored Product, Display, Shoppable Display, and Shoppable Video formats. Instacart works with more than 5,500 CPG brands according to PPC Land's coverage of that accreditation. The platform has also expanded its off-platform data partnerships, including a collaboration with PubMatic that allows advertisers to apply Instacart's first-party retail data to CTV campaigns with closed-loop sales measurement.

European standardization work has proceeded in parallel. IAB Europe updated its pan-European retail and commerce media landscape map in October 2025 and opened public comment on updated Commerce Media Measurement Standards V2, with 78% of stakeholders in IAB Europe research identifying media measurement as requiring industry alignment. The IAB Europe measurement standards process and the IAB's November 2025 incrementality guidelines represent a sustained effort to establish causal measurement as the baseline expectation for commerce media, not an optional enhancement.

The broader commerce media market gives these debates urgency. Commerce media spending is projected to approach $100 billion globally by 2028, according to IAB data. Mastercard entered the space in October 2025 with a commerce media network leveraging transaction data from 160 billion annual payments. Kroger Precision Marketing recently integrated with Google's Display and Video 360 to enable SKU-level conversion reporting on YouTube for CPG brands. These developments expand the pool of closed-loop data available to advertisers - which is precisely what the IAB paper argues should sit at the center of retail media measurement rather than at the periphery.

The paper's conclusion does not dismiss MMM. It states plainly that MMM is not going away. What it calls for is an adoption of best practices and standards for retail or commerce media measurement that define how to measure the true business impact of commerce media investments. Where those facts are already observable, continuous, and verifiable, the paper argues, there is little justification for preferring an estimate.

Timeline

• 1960s: Marketing Mix Modeling introduced as an advertising measurement methodology
• August 28, 2024: IAB Australia releases retail media measurement guidelines leveraging IAB US and IAB Europe frameworks; covered by PPC Land
• March 2024: Google launches Meridian open-source MMM in limited availability; covered by PPC Land
• March 29, 2024: Instacart receives Media Rating Council accreditation for Sponsored Product, Display, and Shoppable formats; covered by PPC Land
• January 29, 2025: Google makes Meridian globally available to all marketers and data scientists; covered by PPC Land
• March 26, 2025: IAB Europe publishes updated pan-European Retail and Commerce Media Definitions
• July 15, 2025: IAB Europe releases updated Retail and Commerce Media 101 Guide and best practice guides; covered by PPC Land
• July 15, 2025: Prescient AI announces first MMM built from scratch since the 1960s; covered by PPC Land
• September 9, 2025: IAB and IAB Europe release incrementality framework for commerce media; covered by PPC Land
• September 30, 2025: Google updates Meridian with pricing variables, channel priors, and binomial adstock functions; covered by PPC Land
• October 1, 2025: Mastercard enters commerce media with network leveraging 160 billion annual payments; covered by PPC Land
• October 9, 2025: IAB Europe opens public comment on Commerce Media Measurement Standards V2; covered by PPC Land
• November 3, 2025: IAB and IAB Europe release Guidelines for Incremental Measurement in Commerce Media; covered by PPC Land
• February 7, 2026: Research finds commerce media underrepresented in 50% of MMM implementations; covered by PPC Land
• April 7, 2026: IAB and Instacart publish "Is Your Legacy Measurement Sabotaging Your Growth in the Retail Media Era?"

Summary

Who: The Interactive Advertising Bureau, represented by Vice President of Commerce and Retail Media Collin Colburn, and Instacart, represented by Measurement Science Manager Priyash Shahane, published the paper jointly.

What: A white paper arguing that Marketing Mix Modeling systematically undercounts the impact of retail media investments due to three structural mismatches - always-on budget patterns that MMM cannot detect, fragmented commerce data that falls outside model inputs, and the availability of superior causal measurement that makes correlational estimation unnecessary for retail media specifically. The paper calls for MMM to be repositioned as a secondary, strategic tool rather than the primary decision engine for retail media budgets, with incrementality testing and closed-loop measurement placed at the center.

When: Published April 7, 2026. The paper builds on IAB's November 2025 Guidelines for Incremental Measurement in Commerce Media and reflects a broader industry debate about measurement methodology that has developed across 2024 and 2025.

Where: The paper is published on IAB's website. It addresses CPG brands and marketers operating across retail media networks in the United States and broader markets where commerce media spending is concentrated.

Why: Commerce media spending is projected to grow 12.1% in 2026 according to IAB forecasts, making measurement accuracy a high-stakes question. The paper argues that brands relying primarily on MMM for retail media decisions are making budget allocations based on estimates when transaction-level, causal evidence is already available from the retail media networks themselves - leading to systematic underinvestment in a channel the paper describes as one of the most effective in the CPG playbook.