Choose Metrics That Help You Act

A dashboard can contain plenty of numbers and still leave a manager unsure what to do. Activity is visible, colors change, and reports arrive on time, yet the conversation remains vague. The problem is often that the measures were selected before anyone defined the decisions they should support. A useful metric helps someone notice a meaningful condition, understand its significance, and choose an appropriate response. Counting alone does not accomplish that.

Start with a management question. Are customers waiting too long? Is work being returned because requirements are unclear? Are projects absorbing more effort than expected? Each question points toward a different measure. Writing the question first prevents a common mistake: treating an available number as important simply because the system can produce it. The measure should serve an operational purpose that the responsible person can explain in ordinary language.

Define what the number includes and excludes. “Completed work” could mean internally finished, approved by a reviewer, delivered to the customer, or accepted by the customer. Those are different events. Agree on the start and end points, the unit being counted, and how unusual cases will be handled. Without shared definitions, teams can spend a review meeting debating the calculation instead of examining the condition the number was supposed to reveal.

Pair output with a relevant quality measure. Completing more requests is encouraging only if those requests are actually useful and correct. A team that responds quickly may still create frustration if the response fails to resolve the issue. Consider what could be sacrificed when people try to improve the headline number. Then include enough context to make that tradeoff visible. The goal is not a larger dashboard; it is a less misleading one.

Look at patterns before reacting to isolated changes. A sudden movement might reflect a genuine problem, an unusual case, a reporting change, or normal variation. Ask what changed in the underlying work. When possible, examine the actual cases behind the number. A measure should direct attention toward investigation, not replace investigation. Management becomes brittle when every fluctuation triggers a new instruction before anyone has established what happened or whether it matters.

Decide who can act on the information. Assigning a metric to someone who cannot influence its causes creates frustration and defensive reporting. Some outcomes require shared responsibility across teams. In those situations, clarify who brings the right people together and who owns the resulting action. Ownership does not mean one person caused every result. It means somebody is responsible for ensuring that the information leads to an appropriate response.

Watch for behavior created by the measure itself. People may avoid difficult cases, split work into smaller counted units, or close tasks prematurely if the metric rewards those choices. These responses do not always indicate bad intent; sometimes they expose a poorly designed target. Ask employees what they would have to sacrifice to improve the number. Their answer can reveal whether the measure supports the business outcome or merely rewards a convenient appearance of progress.

Review the usefulness of the metric as well as the result. If a number repeatedly generates no decision, explanation, or learning, consider removing it. If important problems are appearing outside the dashboard, add a more relevant signal or change the definition. A focused set of measures can support better conversations than an impressive wall of data. Keep the standard practical: does this information help the responsible people understand the work and act more intelligently?