Individual stories, sales calls, or support tickets. Useful for generating questions; insufficient to commit spend against.
The same theme appears across multiple sessions, segments, or sources. Strong enough to justify a scoped investigation.
Same pattern reads true in behavioral data, attitudinal data, and market signal. The finding begins to deserve a product bet.
Finding has survived a pre-registered test with stated guardrails. The result would have changed the team's mind if it had failed.
Evidence has moved from study to production outcome, with post-ship telemetry feeding the next decision's prior.