MIDA / DASHBOARDS & BI

Five questions to answer before your first dashboard

A useful report starts with the decision someone needs to make.

Who is making the decision?

Choose one primary audience. A manager deciding which channel needs attention has different needs from an analyst reconciling fees. Write down two or three recurring decisions and the meeting or workflow where they happen. The first dashboard should support those decisions before expanding to every possible use case.

What does each metric actually mean?

Revenue, active customer and profit need written definitions. Include the formula, exclusions, reporting grain and owner. For example, an average order value based on completed orders should not silently include cancelled orders in its denominator. Definitions need to be visible in the report or its supporting guide.

Can the inputs be trusted?

Check missing dates, duplicated identifiers, incomplete periods and inconsistent currencies. Reconcile one short period with the source report. A visually polished chart cannot compensate for an incomplete export. A visible quality check is more useful than an unqualified green “synced” badge.

Can someone investigate a change?

Provide a route from the headline measure to the relevant channel, period or records. Keep filters consistent across the page and make empty results explicit. The Data Workbench uses this pattern: choose a currency and channel, inspect the filtered rows, then export the records behind the result.

Who keeps the report working?

Name an owner for the source, metric and report. Agree the refresh interval, issue-reporting route and handover documentation. Review whether users can explain a result and take a reasonable next step. A dashboard is successful when it becomes part of a repeatable decision process.