Budgets are shifting toward channels that marketing mix models were never built to register. Commerce media spending is projected to grow 12.1 percent in 2026 and approach 100 billion dollars globally by 2028, according to IAB forecasts. Gaming already draws well over 200 million players in North America alone. Both are getting bigger line items in media plans. Neither shows up cleanly in the models companies use to decide whether that spending is working.
IAB’s State of Data 2026 report puts a number on the gap. Based on a survey of senior planning and analytics decision makers at U.S. brands and agencies, the report found that not a single respondent said their marketing mix model covers all of their paid media channels. Seventy seven percent said gaming is underrepresented, and roughly half said commerce media and the creator economy are missing from their models entirely. A separate 44 percent said they still question the reliability of their incrementality results, the top barrier to measurement confidence the survey recorded.
A follow up IAB paper, co-authored by IAB vice president Collin Colburn and Instacart’s Priyash Shahane, explains why retail media specifically breaks the model. Mix models look for variation over time to detect impact, but retail media spending tends to run always-on rather than in discrete campaigns, so it reads as baseline rather than a performance driver. The paper also points out that most mix models ingest point-of-sale data from physical stores only, missing delivery, pickup, and marketplace transactions. As the paper puts it, the lift from that spending is real, but it has “nowhere to land” inside the model. The paper recommends treating mix modeling as a secondary, strategic tool for retail media and relying on closed-loop incrementality testing as the primary decision engine instead.
Marketers already lean that way, just not far enough to close the confidence gap. A 2025 EMARKETER and TransUnion survey of 500 senior U.S. decision makers found 60 percent trust independent incrementality testing more than mix modeling, which 40 percent trust, or in platform reporting, which 37 percent trust. Yet only 52 percent of marketers were running incrementality tests at all as of that survey. The preference for incrementality is clear. Actually running it, on the channels where mix modeling is weakest, still lags behind.
For consulting teams, the useful question isn’t which measurement stack a client runs, it’s which channels sit outside all of it. A mix model that can’t see gaming or retail media isn’t broken in a way a dashboard refresh fixes. It’s missing the data inputs those channels never generate in a form the model can use. Asking a vendor to name the channels their model excludes, rather than the ones it covers, is a faster way to find out where a budget recommendation is actually guessing.
Sources referenced in reporting: