How to Choose the Right KPIs for a Dashboard
A dashboard earns its place only if every metric on it changes what someone does next. The practical rule is to cap the dashboard at 5-9 KPIs, pair at least one leading indicator with each lagging one, and cut anything that is interesting but not actionable.
Quick answer: Pick 5-9 KPIs per dashboard (Miller's law on working memory is the reason, not superstition). For every lagging KPI (revenue, churn, NPS) include a leading KPI that predicts it (pipeline coverage, support ticket backlog, activation rate). If a metric cannot be tied to a specific decision or owner, it belongs in a report, not a dashboard.
What's the difference between a leading and a lagging KPI?
A lagging KPI reports an outcome that already happened (revenue, churn rate, defect count); a leading KPI predicts that outcome while there is still time to act on it (qualified pipeline, onboarding completion, code review turnaround). Dashboards that show only lagging metrics tell you the score after the game is over.
| Function | Lagging KPI | Leading KPI |
|---|---|---|
| Sales | Closed-won revenue | Qualified pipeline coverage |
| Customer success | Monthly churn rate | Product usage / login frequency |
| Support | CSAT score | Open ticket backlog age |
| Product | Feature adoption % | Trial-to-activation rate |
| Engineering | Production incident count | Code review turnaround time |
How many KPIs should one dashboard show?
Keep a single dashboard view to 5-9 KPIs. That range comes from how many discrete chunks of information people can hold and compare at once; past that, viewers stop scanning and start hunting. If you have more metrics that genuinely matter, split them across a summary view and a drill-down view rather than cramming one screen. See dashboard design principles for how that split should look.
What makes a metric worth putting on a dashboard?
A KPI earns a spot only if it passes four tests: it is actionable, owned, comparable, and timely.
- Actionable — someone changes their behavior based on the number moving.
- Owned — a specific person or team is accountable for it, not "everyone."
- Comparable — it has a target, a benchmark, or a prior period to compare against; a number alone with no reference point is not a KPI, it is trivia.
- Timely — the data refreshes fast enough to matter for the decision it supports. A weekly-refreshed number on a dashboard meant for daily triage is worse than no dashboard at all.
What's a vanity metric, and how do I spot one?
A vanity metric moves in a way that always looks good but rarely drives a decision, most often because it only goes up (total signups, total downloads, cumulative users). Test any candidate KPI by asking "if this number got worse next week, what specifically would we do differently?" If nobody can answer, cut it.
Common mistakes
- Putting every metric stakeholders ask for on one dashboard instead of pushing back on scope.
- Showing only cumulative totals, which can never go down and therefore never signal a problem.
- Mixing metrics with different refresh cadences (real-time next to monthly) without labeling which is which.
- Choosing KPIs the team cannot influence — displaying macroeconomic or seasonal figures as if they were performance metrics.
Pro Tip: Run a quarterly KPI audit: for each metric on the dashboard, log who looked at it and what decision it drove. Anything nobody referenced in three months gets removed, no exceptions. This keeps dashboards from accumulating metrics forever, which is the single biggest cause of dashboard bloat.
← Back to Visualization Tips