Why your dashboards aren't driving decisions
Most BI tools answer questions nobody is asking. The dashboards pile up, adoption flatlines, and leadership quietly drifts back to gut feel. When that happens the instinct is to blame the tool — wrong chart type, bad UX, needs more drill-downs. It's almost never the tool. It's that no number on the screen is attached to a decision anyone actually makes.
Decoration vs. instrument
A metric is one of two things. An instrument is a number someone watches because a decision hangs on it — when it crosses a line, they do something different. Decoration is everything else: true, well-rendered, and inert.
Most dashboards are 90% decoration. Not because the numbers are wrong, but because nobody can tell you what they'd do differently if the number moved. A chart that changes no behavior at any value is a chart you can delete without losing anything but the comforting feeling of measurement.
The three-question test
For every metric on a dashboard, ask:
- Who acts on this? Name a person or role. Not "the team" — a specific owner.
- What decision does it inform? A concrete action: reallocate spend, page an on-call, pause a rollout, call a customer.
- At what threshold? The value or trend that flips the decision from "no" to "yes."
If you can answer all three, you have an instrument — keep it, and make the threshold visible on the chart. If you can't answer even one, you have decoration. Demote it to a detail page or cut it. A dashboard of fifty metrics where six pass this test should become a dashboard of six.
Add the "so what" before the chart
The reason decoration accumulates is that dashboards get built question-first ("can we see revenue by region?") instead of decision-first ("what will we do about a region that's down?"). Flip the order. Start from the decision, work back to the one or two numbers that trigger it, and build only those. Everything else is available on request — it doesn't need a permanent home on a screen people are supposed to act on.
A useful forcing function: make every dashboard tile carry a one-line "so what" — the action it implies. Tiles that can't write their own "so what" are telling you they don't belong.
Fewer numbers, more decisions
The goal of analytics was never more dashboards. It was more good decisions, made faster, with less guessing. Those two things are often in tension: every extra metric is a small tax on attention, and past a point more measurement produces less action, not more, because the signal drowns.
Prune ruthlessly. A short dashboard where every number changes what someone does on Monday beats a comprehensive one nobody opens by Wednesday. If a metric can't name its owner and its decision, it isn't informing the business — it's just watching it.