Name the audience and the action
Decide whether the page serves an operator, a manager, or someone investigating an exception. An operator may need a queue of stalled items; a manager may need to understand changes across a longer period.
Write down the action each major element supports. If a chart cannot be connected to a decision, question whether it belongs on the first screen. Useful information can remain available in a detail view without competing for immediate attention.
Explain the meaning of the number
State the time window, population, and definition behind a metric. “Open requests” is ambiguous if some cancelled requests remain in the count or if the view excludes a team without showing that filter.
Show freshness where it affects interpretation. A current-looking dashboard built from yesterday’s data can cause the wrong response even when its calculations are correct. Make comparisons use compatible definitions and time periods.
Connect the summary to the work
Allow users to inspect the records behind an important count when their permissions allow it. The route from an alerting summary to the affected work should preserve relevant filters and context.
Review the page during a realistic scenario. Can the person identify the condition that matters, understand whether it needs action, and reach the underlying task? If they still export a file to work out what happened, the dashboard may be summarizing the wrong thing.
Keep a record of metric definitions alongside the implementation. When business rules change, update the definitions and explain the effect on historical comparisons. A dashboard is an interface to a decision model, not just a collection of charts.
A practical example.
A support lead reviewing a morning dashboard needs to decide where to assign attention. “Total requests this year” may be less useful than unassigned urgent requests, items waiting beyond an agreed threshold, and work blocked on another team.
For each proposed metric, ask what action changes when the value changes. Include a route from the number to the underlying records so the lead can investigate and act. Define the time window and exclusions beside the metric or in accessible supporting detail. A sudden drop in open requests could mean completed work, a changed filter, or a failed import. The dashboard needs enough context to distinguish those possibilities before someone makes an operational decision.
Put it into practice.
- Connect every prominent metric to a decision or investigation.
- State the time window, meaning, and freshness of the information.
- Let users reach the records behind a summary without rebuilding its filters.