Retail Media Is Breaking Marketing Mix Models.
Here's the Actual Fix.

Retail media is one of the fastest-growing line items in most marketing budgets, and it's commonly measured like an afterthought: either entirely inside the retailer's own walled-garden dashboard, or bolted onto a marketing mix model that was never designed to handle it. In April 2026, the IAB put a name to the problem in a white paper co-authored with an Instacart measurement lead: legacy measurement approaches, including MMM applied without adjustment, are cheating retail media out of its real value (ppc.land, summarizing the IAB paper).

That's a notable admission from an industry body, not a vendor. And it lands at an interesting moment for anyone measuring media across online and offline channels in one model, because the IAB's critique isn't an argument against MMM - it's an argument against measuring retail media apart from everything else.

A channel too big to measure like an afterthought

The scale is the point. The IAB's own paper forecasts 12.1% growth in commerce media spend for 2026 (ppc.land). That's a channel large enough, and growing fast enough, that treating its measurement as a bolt-on to whatever stack already exists stops being a reasonable shortcut. Retail media networks have multiplied over the past few years - Amazon, Walmart Connect, Instacart Ads, Kroger Precision Marketing, Albertsons Media Collective, and dozens of smaller retailer networks all now sell ad space with their own reporting layer attached. Marketers running a handful of these networks, each with its own dashboard and its own definition of "attributed sales," have ended up with a measurement environment that looks a lot like the walled-garden problem social and search platforms created a decade earlier - just newer and growing faster.

What IAB is actually arguing

The specific critique, per the IAB paper, is about data granularity. MMM was built for media types - TV, radio, OOH, display - that never came with transaction-level data attached. It works by relating aggregate spend to aggregate outcomes over time. Retail media is different: it's sold and reported inside platforms that already have granular, SKU-level sales data. Folding retail media spend into an MMM built for channels that lack that granularity, without adjusting for the fact that retail media sits closer to the point of purchase than a TV ad does, risks misattributing sales that would have happened anyway to the retail media line item - or the reverse, undercounting retail media's actual contribution because the model smooths over granularity it didn't need to lose.

It's worth being precise about what this critique is and isn't. It is not an argument that MMM is the wrong tool for retail media. It's an argument that retail media needs to be modeled with the same rigor applied to every other channel, rather than either left outside the model entirely or dropped in without adjustment.

The earlier case for including retail media at all

This isn't a brand-new debate. A WARC and Amazon Ads report had already argued that MMM, done properly, can cover both traditional media and retail channels in the same model, and that doing so is what reveals the split between a retail media campaign's brand-building effect and its pure sales-activation effect (mediaweek.com.au, summarizing the WARC/Amazon Ads report). That distinction matters commercially: a retail media placement that looks purely like a bottom-of-funnel sales driver in the retailer's own dashboard may actually be doing some of the work a mid-funnel channel does - but only a model that includes retail media alongside everything else can show that.

Retailers are already moving toward standardized measurement

Some of the biggest retail media networks are responding to exactly this pressure. Albertsons Media Collective launched a measurement API and partnered with TransUnion in January 2025 to offer TruAudience marketing mix modeling, specifically pitched as a way to compare retail media performance against other channels and other retailers on common ground (businesswire.com, emarketer.com). That's a retailer effectively acknowledging that its own walled-garden numbers aren't sufficient on their own - advertisers want to know how retail media performed relative to TV, paid social, and everything else in the plan, not just relative to its own platform's prior quarter.

That's the right instinct, but a retailer-provided MMM integration still measures retail media primarily through that one retailer's lens. A brand running media across Amazon, Walmart, Instacart and three other retail networks, plus TV, OOH and digital, doesn't actually solve its measurement problem by adopting five different retailer-specific MMM tools that each cover their own slice. It just replaces one set of walled gardens with a more sophisticated set of walled gardens.

The fix isn't a retail-media-specific model - it's one model

The throughline across the IAB critique, the WARC/Amazon Ads argument, and the retailer measurement APIs is the same: retail media needs rigor, and it needs context. Rigor means accounting for the granularity and purchase-proximity that make retail media different from a billboard. Context means retail media's contribution can only be properly understood sitting next to the TV, OOH, radio, print, paid social and search spend it's interacting with - because retail media rarely creates demand from nothing; it often captures demand that offline and upper-funnel digital media already built.

That's the argument for one causal model spanning every channel a brand actually uses, rather than a retail-media specialist model handed off from a MTA or incrementality-testing vendor, sitting next to a separate offline MMM, sitting next to whatever the retailer itself provides. Every seam between separate measurement systems is a place where double-counting or under-counting can hide, and retail media - sitting at the literal point of purchase, downstream of months of brand-building spend elsewhere - is exactly the channel where that seam does the most damage.

What this means heading into 2027 budget planning

As retail media keeps growing at double-digit rates, the brands that will make confident budget trade-offs between retail media and everything else are the ones who can already see retail media's true incremental contribution inside the same model as TV, OOH and the rest of digital - not the ones waiting for each retailer's own measurement team to tell them how their channel performed in isolation.

The bottom line

IAB's critique is a useful correction, not a reason to abandon MMM for retail media. The real failure mode isn't using MMM for retail media - it's measuring retail media in its own silo, whether that silo is the retailer's dashboard or a standalone retail-media model bolted onto an otherwise offline-only MMM.

Animo was built to avoid that seam entirely: one always-on causal model that includes TV, OOH, radio and print alongside retail media, paid social, search and every other digital channel, so retail media's contribution is measured next to the offline spend it's actually interacting with - not estimated in isolation and reconciled after the fact. If your retail media reporting and your offline MMM are telling two different stories about what's driving sales, that's worth a conversation before the next budget cycle.

Sources

• IAB critique of legacy retail media measurement, April 2026 paper co-authored with Instacart measurement lead, including 12.1% 2026 commerce media spend growth forecast (PPC Land): ppc.land
• WARC & Amazon Ads report on MMM covering media + retail channels and brand/activation split (Mediaweek): mediaweek.com.au
• Albertsons Media Collective measurement API + TransUnion TruAudience MMM partnership, Jan 2025 (Business Wire): businesswire.com
• Albertsons moves toward retail media measurement standardization with API + TransUnion partnership (eMarketer): emarketer.com