Every dashboard accumulates metrics over time. Someone asks for a chart, it gets added, and it rarely gets removed — even after the question that prompted it is forgotten. The result is dashboards that are technically comprehensive and practically useless, because the signal is buried in noise.
The test for whether a metric belongs
A useful filter: would anyone change a decision based on this number moving? If a metric could double or drop to zero and nobody's next action would be different, it's not earning its place on a dashboard — it might still be worth tracking somewhere, but not on the view your team looks at daily.
Metrics that usually pass the test
Metrics tied to a specific action. Conversion rate at a specific funnel step, response time to support tickets, churn in the last 30 days — these are metrics where a bad number has an obvious next step attached to it.
Metrics with a known target. If your team has agreed on what "good" looks like for a number — a target CAC, a target retention rate — that number belongs on the dashboard, because it's doing the job of telling you whether you're on track.
Leading indicators, not just outcomes. Revenue is an outcome. Trial-to-paid conversion rate, or activation rate in the first week, are leading indicators that give you time to react before the outcome metric moves.
Metrics that usually don't
Raw totals without context. "Total pageviews" or "total signups" as a standalone number rarely drives a decision — it's the trend, the rate, or the comparison to a prior period that matters, not the raw count sitting alone.
Vanity metrics with no downstream effect. Social media follower counts, app store ratings, or press mentions can be nice to know, but if they don't connect to a metric your team actually acts on, they belong in a separate report, not the main dashboard.
Metrics nobody can explain. If a chart is on the dashboard and nobody currently working there remembers why, that's usually a sign it survived by inertia rather than by being useful.
A lightweight audit process
Once a quarter, walk through your main dashboard with the team that uses it and ask, for each widget: has anyone looked at this in the last month, and can anyone say what action they'd take if it moved sharply? Anything that fails both isn't necessarily wrong to track — it's just in the wrong place. Move it to a secondary view, or drop it, and keep the primary dashboard reserved for numbers people actually use to decide something.