Product Discovery and UX Audit: Cutting a Roadmap by 60% to Ship Faster

Behavioural analysis before opinions: six months of usage data across all 1,100 users. The finding reframed everything — 71% of total in-product time was spent in four screens out of sixty-two. Eleven of the 31 roadmap features were extensions of screens used by fewer than 4% of users; two were extensions of screens with effectively zero usage — features built for a single vocal customer, generating maintenance cost and nothing else.

Contextual observation: Envion observed 23 users doing real work in seven facilities — screen recordings and task timing, not interviews about preferences. The "hard to use" complaint was not distributed across the product; it was concentrated in one workflow — shift handover — where a task users performed 8–14 times daily took 4.5 minutes, five screens, and re-entering data the system already held. Nurses at four of seven sites had built parallel paper or spreadsheet systems for handover, then transcribed into the product afterward. The product was being worked around, not worked in — a signal that never appears in a feature request.

Churn correlation: churned accounts cross-referenced against usage patterns showed churn correlated strongly with time-to-first-value in onboarding and with handover workflow abandonment — and with none of the feature gaps cited in competitor comparisons. The UX audit of the four high-traffic screens found 34 issues, nine causing measurable task failure — including a save-confirmation pattern users misread often enough to lose data, explaining a support ticket category the team had classified as user error for two years.

The seven-week engagement covered product discovery, the UX audit, behavioural analysis and roadmap strategy — usage-data analysis, contextual observation in seven facilities, churn correlation and the heuristic and accessibility review.

Practical rules from the project: your usage data already knows — look at where time is actually spent before any research, and most teams find a version of the 71%-in-four-screens result, which costs a day; watch people work, don't ask them what they want — users describe preferences accurately and behaviour poorly, and they will never volunteer the workaround they have normalized; correlate churn with behaviour, not anecdote — "we lost them because we lack feature X" is usually the tidiest available story rather than the true one; and treat deprecation as a discovery outcome — the features you decide not to build return capacity permanently.