The systems marketers use to prove their budgets work are failing, and the bill for fixing them is enormous. The IAB’s State of Data 2026 research — produced with BWG Global, sponsored by Dstillery and OptiMine, and based on more than 400 senior planning and analytics decision-makers at U.S. brands and agencies — found that between 67 and 76 percent of buy-side decision-makers use incrementality testing, attribution analysis or marketing mix models, yet up to 75 percent say those core approaches underperform on coverage, consistency, timeliness and trust.
The IAB estimates that up to $26.3 billion in media investment value could be unlocked if measurement systems were fixed first, allowing AI to deliver insight faster and more strategically; related coverage of the same research puts the potential value of AI-powered measurement as high as $32 billion. Either way, the mechanism is straightforward: automated buying systems now allocate real money in real time, and they allocate it according to the signals they are fed. Broken signals do not produce neutral results — they produce confidently wrong budgets at machine speed.
The causes are structural. Privacy regulation and platform changes have scattered data across disconnected systems; AI-mediated search and zero-click experiences have made journeys partially invisible; and the models themselves have blind spots — the research found 77 percent of marketers acknowledging that gaming is underrepresented in their mix models, half saying commerce media is overlooked, and more than four in ten flagging connected TV as missing. Boards, meanwhile, are shrinking marketing’s share of company spending, which makes every measurement failure more expensive than it was when budgets were forgiving.
The response inside organizations is becoming contractual. More than a third of buy-side teams have already added AI governance clauses to vendor contracts covering transparency, security and model accountability, and that share is expected to double by 2027. Half of marketers anticipate legal, privacy or accuracy challenges with AI-driven measurement within two years.
For marketing leadership, the sequence matters more than the tool. Fixing measurement before scaling AI buying is not caution; it is arithmetic. An optimization engine is a lever that multiplies whatever truth — or error — it is given. The organizations that treat measurement infrastructure as a capital investment, rather than a reporting cost, are the ones whose automation compounds value instead of waste.
The study’s underlying warning travels well beyond media planning. Any organization that hands budget authority to automated systems before repairing the evidence those systems consume has not modernized its marketing; it has automated its uncertainty, and it will pay for the privilege at full price.
Related reading: AI Agents Move Into the Marketing Stack: Breeze, Agentforce and the End of the Blank Field
