Most corporate dashboards carry between twenty and forty indicators. In practice, people look at three. The rest exist because someone asked for them once, because the data happened to be available, or because they filled a gap on a slide.
This is not a dashboard design problem. Those indicators should never have made it in.
The four questions
An indicator earns its place when it passes all four. Fail one and it’s dead weight.
Can someone act on it this week? If the answer is “it depends on the market” or “headquarters decides that,” it isn’t a management indicator — it’s context. Context belongs in the monthly report, not on the operating dashboard.
Does the action move the number within a visible timeframe? Cycle time responds in days. Customer satisfaction responds in months. Putting both on the same dashboard with the same review cadence turns one of them into noise.
Is there a name attached? Not a department: a person. “Operations” is accountable for nothing. If nobody feels uncomfortable when the number turns red, the number isn’t going to move.
What specific decision changes with the value? This is the question that kills the most indicators. If you can’t finish the sentence “if this number goes past X, then we do Y,” what you have is a reporting metric, not a KPI.
The signs it’s about to be ignored
A few patterns predict fairly well which indicator nobody will be looking at in three months.
It’s an average with no dispersion. Average handling time can be 4 minutes with every case between 3 and 5, or with half of them at 1 minute and half at 8. Those are two different operations and the average shows them as identical. An average without its 75th percentile or its standard deviation hides more than it reveals.
It’s a percentage without the denominator. “94% compliance” over 50 cases and over 50,000 cases call for different conversations. The denominator goes right next to it, always.
It measures activity instead of outcome. Number of meetings, reports produced, tickets opened. These go up when the team works harder, not when the business improves. They’re the first to get inflated when someone feels watched.
Nobody knows where it comes from. If explaining the figure requires opening three files and asking a colleague, the indicator has already lost. Trust in the number is part of the number.
An exercise that works
Before building the dashboard, sit down with whoever is going to use it and ask them to complete this sentence for each proposed indicator:
When I see that [indicator] hits [threshold], I will [specific action] and I expect it to change within [timeframe].
The indicators that can’t complete the sentence fall away on their own — no argument, and nobody has to defend their turf. In a typical run, five to eight survive out of the thirty that were on the list.
And a dashboard with eight indicators is one people actually look at.