Your Customer Went Quiet, and Nobody Noticed

The problem

A software company tracked customer health mostly through support tickets — if an account wasn’t filing complaints, it was assumed to be fine. But a ticket-only view has a blind spot: an account can stop using a product entirely without ever telling anyone why.

The question

Which accounts were actually at risk — and were the noisy, ticket-heavy accounts the same ones that were actually declining, or a different group entirely?

The approach

A second, independent view of account health was built directly from product usage data, running in parallel with the existing support-ticket analysis. The two signals were then compared account by account, sorting each into one of four patterns: declining on both usage and tickets (most urgent), declining on usage while staying quiet on tickets (the blind spot), noisy on tickets while usage stayed flat (a trust problem that hadn’t hit the numbers yet), or no real movement at all.

What I found

Several accounts fell into the second pattern — usage quietly collapsing, with no ticket ever filed. A ticket-only view would have missed every one of them until the account was already gone.

What changed

The two-signal comparison became a recurring monthly report, replacing a fully manual process and giving the team a way to catch disengagement before it turned into churn instead of after.