The dashboard was supposed to democratise data. Mostly it created a new bottleneck: a wall of charts nobody quite trusts, and a data team fielding the same five questions in Slack because the answer is "close but not exactly" what the chart shows.
The question is the interface
People do not think in charts; they think in questions. How much did that campaign really make us? Which customers actually stay? The right interface lets someone ask that in plain language and get a direct answer, not a link to a dashboard they then have to interpret.
Trust comes from transparency
An answer you cannot inspect is just an opinion. So every answer should carry its receipts: the metric definition, the filters applied, the date window, and the query that produced it. When a marketer can see exactly how a number was built, they stop pinging the data team to double-check, and the data team gets its afternoons back.
- Plain-language questions map to catalog metrics and dimensions, not ad-hoc SQL guesses.
- Every result exposes the query and filters behind it.
- The same number appears identically wherever it is shown, because it flows through one code path.
For the first time our teams argue about strategy instead of arguing about whose number is right.
Dashboards still have a place
None of this kills the dashboard. A well-built board is still the best way to monitor something over time. But the default should flip. Start with the question, get the answer, and let the dashboard be where answers you keep asking go to live.