Reports people can trust.
WHAT MAKES A DASHBOARD FAIL
Numbers that disagree with other sources Stale data with no indication Slow loading Too much shown at once No clear purpose
WHAT TO ESTABLISH FIRST
What decision the dashboard supports, and who makes it.
WHY
A dashboard supporting no decision is built, admired, and abandoned.
WHAT TO SHOW
The few numbers that matter, and what they should be compared against.
WHY COMPARISON MATTERS
A number alone tells nobody whether it is good.
WHAT TO ALWAYS DISPLAY
When the data was last updated.
WHAT TO ADD
Definitions, reachable from the dashboard itself.
WHY
The first question about any figure is what it includes.
WHAT TO BUILD ON
Modelled tables, not raw data or ad hoc queries.
WHAT TO AVOID
Logic embedded in the dashboard Filters that silently exclude data Many dashboards covering the same ground
WHAT TO REVIEW
Usage, and retire what nobody opens.
WHY
Unused dashboards still cost money to refresh and confuse people who find them.
WHAT TO TEST AFTER EVERY MODEL CHANGE
That the dashboard still shows the same figures.
