A headline circulating in trade press from Roastbrief US reports that 97% of global marketers admit their organizations are not ready to fully harness AI for marketing measurement. The figure, presented as a single statistic from an unspecified survey instrument, has become a shorthand reference point for conversations about the gap between AI ambition and AI execution inside marketing organizations.

What does "not ready" actually mean?

Marketing measurement has long combined three components: data collection, attribution modeling, and reporting that informs budget decisions. AI tooling now sits across all three — from automated tag management and server-side event capture to model-based attribution and predictive forecasting. The headline's framing implies respondents are being scored against the full stack rather than against a single tool installation. That distinction matters. A team that has purchased an AI attribution platform is not the same as a team that has rebuilt its data layer to feed the model and wired the output into its media planning workflow.

How to read the 97% number

Self-assessed readiness surveys in marketing tend to land in a predictable pattern. Respondents report piloting AI somewhere in their stack at high rates, and report full end-to-end deployment at much lower rates. A finding that puts 97% on the "not ready" side sits at the most pessimistic end of that range. Two factors usually push such numbers higher. First, respondents tend to benchmark themselves against an aspirational target rather than current capability. Second, they often conflate "an AI tool is installed" with "the AI tool is driving decisions."

The operator's problem underneath the headline

For many marketing teams, the binding constraint is not the AI model itself. It is the data layer feeding the model. First-party data fragmented across CRM, point-of-sale, web, and app events produces attribution models that overweight last-click channels and underweight brand activity. AI-assisted measurement is meant to address that pattern — but only if the underlying data is structured and reconciled. Many marketing organizations have not completed that work, which is consistent with the gap the headline describes.

What the survey instrument likely does not show

The headline carries no detail on methodology, sample composition, geographic distribution, or how "ready" was defined. A 97% finding is striking, but operators should treat it as a directional signal rather than a measured baseline. Without the underlying survey instrument, the number cannot be benchmarked against prior waves or against peer industries running similar exercises.

A reasonable next move

Marketers evaluating their own stack can score readiness against the three layers above — data, model, and decision workflow — rather than against vendor demos. The honest internal score usually lands closer to the headline's 97% than to a vendor pitch deck. Operators who want to close the gap should start with the data layer, since no amount of model sophistication compensates for unstructured data underneath.