The three questions every dashboard should answer

Every dashboard project I have worked on started with the same request: show everything. The data is there, the team is proud of it, and nobody wants to be the one who hides a number. Six weeks later the dashboard ships, people open it twice, and then they go back to exporting spreadsheets.

The fix is rarely visual. It is a short list of questions, written down before anyone opens a design tool.

Start with the morning

I ask customers one thing in the first interview: what do you check before your first meeting? The answers are surprisingly consistent. At Northwind, twenty people gave me three questions. Is anything broken? Are we on track for the month? What changed since yesterday? Everything they actually did in the product was a version of those three.

Those questions become the top of the screen. Not as tabs or filters, but as three plain answers with a number, a direction and a one-line reason. If the answer is boring, the dashboard is working.

Move everything else one click deeper

The hard part is not adding the three answers. It is deciding where the other forty charts go. My rule is simple: if a chart doesn’t help answer one of the three questions, it moves one click deeper, into a report people open on purpose. Nothing gets deleted, so nobody has to fight for their metric in a meeting.

This also makes the design easier to maintain. New metrics have an obvious home, and the first screen stays calm even as the product grows.

Test with real mornings

Prototype reviews in a meeting room lie. People are polite, they read every label and they have time. Instead I send a clickable prototype to five customers and ask them to use it at 9am, before their stand-up, for one week. The questions they still ask in Slack are the ones the dashboard failed to answer.

At Northwind that week found two problems we would never have seen in a review. Both were fixed with copy, not with charts. Daily active use grew 64% in the three months after launch, and exports fell by half.

If you are about to design a dashboard, write the three questions first. If you can’t, the project isn’t ready for design yet, and that is worth knowing early. You can read how that project went in the Northwind case study.