Home › Guides › KPI dashboard

How to build a KPI dashboard

A few measures people act on beats forty nobody reads.

Dashboards fail by including everything measurable. The discipline is choosing what to leave out.

Step-by-step

  1. Choose five to eight measures.
  2. Set a target for each, noting where lower is better.
  3. Enter actuals each period.
  4. Read the variances and act on the exceptions.

Choosing what to measure

A useful measure has three properties: somebody can influence it, it is available soon enough to act on, and it connects to something that matters. Anything failing all three is decoration.

Mix leading and lagging indicators. Revenue tells you what happened; enquiries and pipeline tell you what is about to. A dashboard of only lagging measures reports the past accurately and always too late.

Targets, and lower-is-better

Some measures improve as they fall — churn, defect rate, days to collect. A dashboard treating every increase as good will mislead exactly where it matters most, so mark those explicitly.

Measures become targets, and then stop working

Anything measured and rewarded gets optimised, including in ways you did not intend. Measure calls handled and calls get shorter, not better. Measure tickets closed and they get closed prematurely.

The defence is pairing: alongside volume, measure quality. Alongside speed, measure rework. A single measure in isolation will eventually be gamed, usually without anybody deciding to.

If nothing on the dashboard has ever changed a decision, it is a report rather than a dashboard. Cut it to the measures that would actually make you do something differently.

Frequently asked questions

How many KPIs should I track?

Five to eight at any level. More than that and attention is spread too thin for any of them to drive action.

What is the difference between leading and lagging indicators?

Lagging measures record outcomes, like revenue. Leading measures predict them, like enquiries or pipeline. You need both, or you learn about problems only after they have happened.

Why do measures stop working over time?

Because what is measured gets optimised, including in unintended ways. Pair each measure with a counterbalancing one — volume with quality, speed with rework.

Open the decision tools →