AI-assisted reporting records significantly fewer errors than manual workflows, yet 65% of marketing leaders still won't share AI-generated insights without further validation

Marketing leaders are almost twice as likely to trust a report produced without AI than one generated by it, even though manual reporting is more prone to error.

New research from marketing data intelligence company Adverity found that 64% of marketers are comfortable sharing AI-free data insights without additional checks; however, this falls to just 35% when those insights are generated using AI.

The figures point to a wider problem with AI adoption in marketing. The technology may be producing more accurate outputs, but marketers do not necessarily trust those outputs enough to act on them.

Nearly a quarter (22%) of manual reporting workflows experience regular errors, compared with 6% of automated processes and just 2% of AI-assisted reporting. Yet the greater accuracy of AI-assisted reporting is not translating into confidence, with many marketers still reluctant to put those insights into circulation without further scrutiny. That additional review layer undermines the time from insight-to-action that many are looking to benefit from by implementing AI in the first place.

The research highlighted that much of that caution comes down to missing business context (39%). An AI model can identify a change in performance or flag an anomaly, but without a clear understanding of how a business defines its metrics, what is driving changes in campaign performance (such as budget, promotions or live tests) and what teams are defining as success, marketers are left checking whether the answer makes sense before they can use it.

That challenge becomes particularly clear when even internal teams cannot agree on what their own data means. 94% of marketing departments say they do not consistently use the same metric definitions across the business, while 82% say data disputes have delayed major decisions in the last quarter. This underlines an important finding: if people across an organisation cannot work from a shared definition of a metric, an AI system has little chance of reliably interpreting it.

This gap between experimentation and ROI explains why so many AI initiatives stall early. While AI has the capacity to drastically reduce time-to-insights, high adoption rates are not translating into real payoffs. Without sufficient organisational context, AI models inherit existing data inconsistencies, producing authoritative-sounding hallucinations that leave marketers bogged down in manual verification.

This fragile trust directly limits how far businesses can scale AI. Teams have seen flashes of AI’s ability to speed up advanced analytics, but confidence disappears the moment an answer is wrong. As a result, the majority of AI investments fail to deliver, with only 2% allowing AI agents to make autonomous optimisation decisions while the rest remain confined to low-risk, back-office tasks.

“AI is only as useful as the information and context it has to work with,” said Alexander Igelsböck, CEO of Adverity. “Marketing teams have good reason to question an answer when they cannot see where it came from or understand why a model reached a particular conclusion.

“The answer is not to put more checks around AI. It is to give it better foundations. A dedicated knowledge layer can connect marketing data with business rules, governance and strategic goals, giving AI the context it needs to produce insights marketers can actually trust and act on.”

Methodology: This study is based on a survey of 300 senior marketing professionals and brand leaders across the UK, US, Germany, Austria, and Switzerland. Respondents span major enterprise sectors - including marketing agencies, retail, and financial services - and hold active decision-making roles such as Chief Marketing Officers, Vice Presidents, and Marketing Directors.