Marketing’s AI Problem Isn’t AI. It’s Data.
435 marketers. 80% under pressure to adopt AI. 6% have managed it. The gap is a data problem.
The Supermetrics' 2026 Marketing Data Report surveyed 435 marketers globally and found something the industry would rather not say out loud: the pressure to adopt AI is overwhelming, but the results are not.
Eight in ten marketers report feeling under pressure to introduce AI into their workflows. Just 6% say they’ve actually managed it. That gap, between mandate and reality, is the report’s central finding, and it points to a problem that new tooling alone won’t resolve.
The pressure is coming from the top. According to the report, 89% of the pressure for AI adoption originates with C-suite executives and boards. The logic is understandable: AI is moving fast, competitors are experimenting, and leadership wants results. But the survey data, collected across brands and agencies in the US, UK, Germany, Australia and Singapore, tells a more complicated story about what’s actually blocking progress.
It isn’t reluctance. Most marketers believe AI could help them, 87% say better data and analytics would improve their team’s effectiveness. The real problem is that the data infrastructure required to make AI useful simply isn’t in place. Over a third of respondents report not having access to high-quality, accessible data. A similar proportion cites a lack of system integration between analytics tools and activation platforms. More than half say they wait 1 to 3 business days for answers from their data teams, and 73% are unhappy with the frequency of that support.
Zach Bricker, North American Solutions Engineering Lead at Supermetrics, is direct about it in the report: “teams are being told to use AI without the support to make it effective. When clear goals are absent, adoption becomes performative rather than functional.”
The ROI problem that won’t go away
Proving return on investment remains the hardest thing marketing teams are asked to do. 45% named it their number one challenge, yet 61% are required to demonstrate it to justify the budget. The irony is that ROI, still the top KPI cited by 45% of respondents, is not actually provable in any precise sense. Attribution modelling is too imperfect, human behaviour too unpredictable, and the variables too numerous for a clean number to emerge.
Linehan’s argument in the report is that budget pressure on marketing is rarely about a business’s financial health; “it’s driven by a deficit of trust in marketing’s ability to demonstrate commercial value.”
Measurement priorities are moving. A/B testing remains the most common approach at 69%, while marketing mix modelling — long considered too resource-intensive for all but the largest advertisers — has become the top investment priority for the next twelve months, cited by 40% of respondents. Agencies are ahead here: 59% already use MMM, compared to 35% of brand-side teams. Incrementality testing, last year’s rising priority, has retreated — perhaps a sign that teams overestimated their readiness to implement it.
The shift toward MMM reflects something real: marketers are trying to construct a more defensible picture of what’s working, one that can survive a budget meeting. Whether they have the data foundation to make it work is a different question.
Who actually owns the data
One structural problem runs through almost every finding in the report. In most organisations, data strategy is controlled by people outside marketing. Over half of respondents said that data decisions, what gets collected, how it gets measured, and what it means, are made by external data or IT teams. Only 31% said the CMO plays a meaningful role.
This matters because the data collected reflects the priorities of whoever controls the collection. When IT or central data teams lead, the result is data that’s efficient to gather and difficult for marketing to act on. As Linehan argues in the report, without marketing involvement in data decisions, you end up with data that’s comprehensive in volume but weak in relevance.
The report draws a useful distinction between four stages of the data workflow: connecting data, managing it, analysing it, and activating it. The argument is that marketing should own activation outright, the translation of insight into campaign decisions, and should be meaningfully involved in connection and analysis. Data management is the responsibility of the engineering and data teams. The problem is that most marketing functions have ceded too much ground across the board, leaving themselves dependent on slow-moving external teams for answers they need quickly.
Too much data, not enough time
45% of respondents say they rarely have enough time to properly analyse what they’re looking at. A third check reports monthly or less. Supermetrics’ own platform data shows a 9% increase in data usage between 2024 and 2025, following a 230% increase in the prior period. Volume isn’t producing proportionate insight.
The companies that derive genuine value from AI in the report share a few characteristics: clear use cases, fixed data foundations, and established habits of acting on data before any AI layer was introduced. The sequence matters. Most organisations are attempting it in reverse.
You can download the Supermetrics 2026 Marketing Report HERE
Very valuable insights, thanks.