Most dashboards delivered with a project are never opened after the first month. Not out of disinterest, but because they show everything, and therefore nothing.
Three indicators, not thirty
We systematically ask the client to choose three numbers before launch: one that measures use, one that measures quality, one that measures economic effect. Three are enough to decide; beyond that, people comment instead of deciding.
The false good measures
Visit counts, accounts created and automation rates have one thing in common: they go up when everything is fine, and also when everything is wrong. A spike in visits can come from an outage shared on social media; a high automation rate can hide customers giving up.
A good indicator is a number that, if it moves, forces you to do something.
Set the threshold before, not after
An indicator without a threshold is useless. Decide before launch at what value you consider it works, and at what value you intervene. Otherwise every number will be read in the direction you were hoping for.
Review every quarter
Good indicators go out of date. The one that mattered at launch — adoption — becomes useless once the tool is adopted. We review every three months and do not hesitate to swap one number for another.
What this avoids
This discipline avoids the most uncomfortable conversation of all: the one where, six months after delivery, nobody can say whether the project was any use.