Most reporting projects start with the data that is easy to get and work forwards, hoping a useful view emerges. The result is a wall of charts that everyone glances at and no one acts on. The numbers are correct; they are simply not connected to a choice.

Work backwards from the choice

A better starting question is not "what can we measure?" but "what decision are we trying to make better, and who makes it?" Name the decision, its owner, and how often it recurs. Only then ask what evidence would actually change it. More often than not, that is two or three numbers, not twenty — and half of what the old dashboard tracked turns out to be there for reassurance, not action.

Fewer metrics, closer to the work

A metric earns its place when someone can point to a time it changed what they did. Everything else is context at best and noise at worst. This is as much an operating-model question as a technology one: a number only drives a decision if someone is genuinely accountable for that decision and trusted to act on it.

The aim is not a more impressive dashboard. It is the smallest set of signals that lets the right person act with confidence — and the discipline to stop measuring the rest.