Data stopped being the bottleneck a while ago. Funnel’s 2026 Marketing Intelligence Report, built from a survey of 238 in-house marketers and agency professionals plus interviews with six industry experts, found that 72 percent of marketers say they already have plenty of data. The problem is converting it into something a team can act on. Funnel’s researchers found the highest performing teams spend less time building dashboards and more time asking why a number moved, which is a different skill than owning more analytics software.
That shows up directly in what marketers want AI to do next. A survey of 110 US marketing managers and above, run by EMARKETER with Rakuten in October 2025, asked which AI feature would matter most for the next generation of marketing mix modeling tools. 60.9 percent picked generative summaries that explain a model’s output in plain language, nearly double the share who picked automated dynamic learning features that let the model retrain and optimize itself. Marketers were offered the chance to hand more control to automation and mostly said no. They want a sentence that tells them why the model moved, not a system that quietly moves itself.
Access to modeling tools is the part that already got solved. Free marketing mix modeling platforms, including Meta’s open-source Robyn and Google’s Meridian, have made it possible for a small brand to build a basic model without buying anything. Industry coverage of the trend, including analysis published by Analytic Partners and reported by MediaPost, points to a split forming underneath that access: directional, do it yourself models on one side, and premium modeling built for scenario planning and forecasting on the other. One case cited in that coverage, a global FMCG company that restructured its planning around scenario based modeling, reported 212 million dollars in realized value in 2024, according to the company’s own account. The tools got cheaper. The judgment required to trust a complex model’s output did not.
Put the three together and the shape of the gap is clear. Marketers have more data and more modeling access than they did two years ago, and both Funnel’s and EMARKETER’s research describe the same unmet need sitting on top of that abundance: someone, or something, that can explain a result credibly enough to act on. Funnel’s own experts put it bluntly, warning that a confident sounding AI answer can still bury the truth underneath it, and that validation matters more now than it did before automation got this cheap. The 2026 marketing analytics stack is not short on horsepower. It is short on a trustworthy explanation for what the horsepower just produced.
That is a different problem than the one most vendors are building for. A product roadmap built around faster model training, more automated retraining, or one more dashboard view does not touch the actual request marketers are making. They are asking for a model that can defend its own answer in a budget meeting, in plain sentences, with the assumptions visible. Until that exists as a default feature rather than a consulting engagement, the split between self serve modeling and the premium version built on top of it is likely to keep widening, not closing.
Sources referenced in reporting: Funnel, The 2026 Marketing Intelligence Report, EMARKETER, Marketers want AI to explain their data, not just crunch it, MediaPost, How Marketing Mix Modeling Is Becoming a Must Have.