Business intelligence dashboards

Reporting dashboards fail for a reason that has nothing to do with charts: nobody agreed what the number means before it was put on a screen. Two business units use the same word for two different measures, the executive view reconciles neither, and the dashboard becomes a thing people screenshot into a spreadsheet to fix by hand.

So the design work ran end to end — from stakeholder research through to tested prototypes — and the earliest part was definitional rather than visual.

Two audiences, two jobs

Business-unit reporting and executive reporting look similar and are not. A business unit needs to find the cause of a movement; an executive needs to know whether a movement warrants attention. Designing one artefact for both produces something that is dense enough to be unreadable and shallow enough to be useless.

• Business unit view — drill-down, comparison, and enough granularity to answer "why".
• Executive view — a small number of measures, consistent framing, and an explicit threshold for what counts as a signal.

Testing prototypes with the actual reporting audience — not proxies — was what separated the measures people used from the ones they had asked for in a workshop.