Systems / 1 min read
A dashboard needs a decision.
Start with what someone needs to decide, then choose what to show.
Design backward from action
Before I sketch a dashboard, I want to name the decision it supports. Is someone choosing where to investigate, whether to release a change, or which work to prioritize? Each question asks for different evidence. A large collection of available metrics can obscure that purpose.
I would pair each important signal with its definition, its freshness, and the action it can inform. If the reader cannot tell what a number means or what changed, another chart is unlikely to help. A useful dashboard should shorten the path from observation to a considered decision.