Every SaaS team I've worked with has a dashboard nobody opens except before the quarterly review. It's usually beautiful. It's usually useless. The tells are always the same: a wall of charts, none of them tied to a decision anyone actually has to make this week.
The metrics that earned their place on my roadmap were never the ones with the most impressive names. They were the boring ones that answered a specific question someone was already arguing about.
The three that kept showing up
Time-to-first-value. Not signups, not activation in the abstract sense vendors sell you — the actual number of minutes or days between someone creating an account and doing the one thing that made them come back. Once we started tracking this against specific onboarding changes, half of our backlog reprioritized itself.
Feature adoption decay. Everyone tracks the spike after launch. Almost nobody tracks the six-week-later number, which is the one that tells you whether you built something people wanted or something they tried once out of curiosity.
Support ticket clustering by feature area. Not ticket volume — volume just tells you where the confusion is loud. Clustering by root cause told us where the confusion was structural, which is a roadmap item, versus where it was a one-off, which is a documentation fix.
A metric earns a place on the dashboard by having already changed a decision once. If it hasn't, it's decoration.
What I stopped reporting
Vanity totals — cumulative signups, total page views, aggregate "engagement" scores that blend five behaviors into one number nobody can act on. They read well in a slide. They've never once told me what to build next.
If you're building your own reporting cadence, the test I use now is simple: for every metric on the page, ask what decision it would change if it moved 20% in either direction. If nobody in the room can answer that, it doesn't belong on the dashboard — it belongs in an appendix, if anywhere.