The measurement trap
There is a seductive version of being data-driven that feels rigorous but isn't. We had too much data and had never done the hard work of deciding which numbers were actually predictive of the outcomes we cared about.
Metrics we stopped using
Page views and unique visitors: they measure interest but not engagement. Signup count: easy to inflate. Social media follower counts: genuinely useless for anything except vanity. Time on page: often means someone was confused, not engaged.
Metrics we kept
Week-two retention: the percentage of new users who return and do something meaningful in their second week. This is the most predictive single number we've found for long-term product health.
“If you only track one number, track week-two retention. Everything else is downstream of it.”
Revenue per active user, MRR with churn context, and Net Promoter Score surveyed every quarter — not automatically, because automated NPS surveys get ignored.
The qualitative layer
Numbers measure what happened. They rarely explain why. We rely on three things: reading every support ticket personally on rotation, watching session recordings for new users who don't return, and scheduling one user interview per week.
Our simple dashboard
Our current weekly dashboard fits on a single screen with eight numbers: week-two retention by cohort, MRR with week-over-week change, active users, churn rate, NPS, support ticket volume, average ticket resolution time, and number of users interviewed that week.
Enjoyed this article?
Get new articles, product updates and insights from Muqira in your inbox.
Found this useful? Share it.
