August 1, 2026
The Data That Sabotages Your Product Decision
29% of employees admit to sabotaging their company's AI strategy. What this does to your satisfaction surveys and feature tests.
A team ran a new feature for two weeks. Almost every evaluation came back with a middling score: not good enough to approve, not bad enough to raise a flag. The feature was cut. But should it have been?
The sabotage signal rarely arrives as an obvious breakdown. It comes as the lukewarm score that changes nothing either way.
The scenario is fictional, but not by much. A study by Writer, in partnership with Workplace Intelligence, surveyed 2,400 employees and executives in the United States, the United Kingdom, and Europe. 29% admitted to sabotaging their own company's AI strategy, and among Gen Z the number climbs to 44%. Cited methods include refusing the company-chosen AI tool and manipulating performance evaluations to make AI look worse than it is. That second one is the real-world version of the lukewarm score at the start.
This data affects your satisfaction survey, your test reports, and every meeting where you approve or cut a product improvement.
76% of executives say this sabotage represents a serious threat to their company's future. From a product leadership perspective, finding who sabotaged matters less than finding the problem and building the fix, because every result is the responsibility of whoever led the process. No single result should decide alone. Qualitative research is not immune to bias either: a leading question already has its answer baked in. Cross-referencing survey scores with what the market shows, with direct user complaints, and other signals helps locate the discrepancy. If a competitor launched the same feature and is seeing results, the problem may be in the pilot or in the metrics being used. And remember: polite silence in the meeting can be an objection waiting for a better moment.
Trust in the process is what sustains trust in your product.