Designing a Dashboard

Farah's first operations dashboard had 22 charts, eight filters and a table of every order. It had everything, and she stopped opening it after a week. A dashboard is not a place to store all the charts you could make. It is a tool that answers one audience's few recurring questions at a glance.

The rule

One audience, a few questions, the most important top-left.

• One audience. A dashboard for Farah differs from one for Meera or for a shift lead. Designing for everyone designs for no one.
• A few questions. Write the three or four questions it must answer before you draw anything. Every visual must answer one of them, or it goes to a detail page.
• The most important top-left. Readers of left-to-right languages scan a screen roughly from the top-left, across, then down. Put the headline numbers there, trends below them, and breakdowns and detail at the bottom.

The headline numbers go in KPI cards. A bare "90%" means nothing, so each card carries context:

• the value: On-time delivery 90%
• against target: ▼ 5 pts below 95% target
• against last period: ▼ 2 pts vs February (92%)
• a small trend line of the last 12 weeks

Keep the rest consistent: the same colour always means the same thing, units are always shown, and chart titles say what is plotted. White space is not wasted space; it groups related visuals.

Many users will open the dashboard on a phone. Design that view on purpose: one column, KPI cards first, then the single most important chart, with detail moved to its own page. Anything shown only on mouse hover is invisible on a touchscreen.

The shortcut

Use the five-second test. Show the dashboard to its user for five seconds, then hide it and ask: "Are we on track, and where is the problem?" If they cannot answer, simplify until they can.

The trap is the data dump: adding a chart for every request until nothing stands out. Say no politely, or give the extra charts their own page.

Drills

Answer fast — trust the rule.

Remember

• One audience, a few questions, the most important top-left.
• Five-second test: "Are we on track, and where is the problem?"
• Trap: the data dump, and KPI cards without a comparison.