Dashboards are the most requested and least used artifact in enterprise software. They get commissioned with enthusiasm, built comprehensively, demoed proudly, and then opened twice a quarter — usually to prepare a slide.
The failure is almost always the same. The dashboard was designed around the data that was available rather than the decisions that get made.
Start from the decision
A useful way in: ask what someone will do differently depending on what the screen says. If the honest answer is 'nothing, but it's good to know', that's not a dashboard — it's a report, and it belongs somewhere else.
Once the decision is named, the design almost falls out of it. A dashboard for 'which jobs need me today' looks nothing like one for 'is this quarter on track'. The first wants a queue, a sort, and an action on every row. The second wants a trend, a comparison, and no actions at all. Built as one screen, both are compromised.
If nobody changes what they do because of a number, the number doesn't need a chart. It needs deleting.
Rules that keep a dashboard honest
- One question per view. A screen that answers four questions answers them all vaguely.
- Every number needs a comparison. A figure alone is trivia; against a target, a previous period, or a peer, it becomes information.
- Say when the data is from. A stale dashboard that looks live is worse than no dashboard, because someone will act on it.
- Make the exception findable, not just visible. If the point is to catch the outlier, sorting and filtering matter more than the chart.
- Let the row be actioned. The distance between noticing something and doing something about it is where most dashboards lose their users.
On chart choice
Chart selection gets more attention than it deserves, but two habits are worth holding. Use position and length before color and area, because people read them more accurately. And reserve color for meaning — a status, a threshold, a series that matters — rather than decoration. A chart where every series has its own bright color is a chart that has declined to say what's important.
Write the sentence a user would say after looking at the screen: “Right, so I need to ____.” If you can't finish that sentence, you don't have a dashboard design yet — and no amount of chart work will produce one.
Working on something this touches? We're usually happy to talk it through before anyone commits to a project.
Start a conversation