Dashboard Design: Fewer Charts, More Decisions
A dashboard earns its place when it helps people decide. How many metrics to show, what goes first, how to use filters and drill-down, and how to split views.
What problem does a dashboard solve?
It summarizes information so someone can act without opening five reports. Its value is prioritization: what gets looked at first, which figure signals a problem, and what happens when something changes. If everything carries the same weight, nothing does.
Girard Media proposes a simple test for each metric: if it doesn’t lead to an action, remove it (Girard Media, “Dashboard design best practices”). It is a good filter before you add any chart.
Why should a designer care?
Because dashboards are among the densest screens you will design, and the worst decisions often come from trying to satisfy everyone who asks for a metric. A dashboard with eighteen charts has no hierarchy, even if each chart is well made. Designing here means deciding what is not shown.
A dashboard is also read quickly, with little interaction. Nielsen Norman Group defines dashboards as collections of data visualizations presented in a single-page view and recommends designing them for at-a-glance reading. To compare quantities it suggests bar, line, or scatter charts, because length and 2D position are the attributes people perceive most accurately. It advises avoiding pie charts most of the time and avoiding 3D effects, and not using color to communicate magnitude (Nielsen Norman Group, “Dashboards: Making Charts and Graphs Easier to Understand”). That makes the problem as much about content as about visual design.
How many metrics, and in what order?
FusionCharts proposes this common structure, from top to bottom:
- KPI cards at the top. Three to five primary metrics tied to business goals, with a large figure and, where possible, a comparison with the previous period.
- Trends. Line charts that show whether the figure is going well or badly over time.
- Breakdowns. By channel, region, or product, to understand what moves the figure.
- Supporting tables. For anyone who needs the exact number, at the end.
For the primary KPIs, Girard Media recommends “usually 3 to 5 metrics” tied to business goals. FusionCharts lists hiding important insights below the fold and displaying too many KPIs as mistakes (FusionCharts, “Dashboard design best practices”). Eleken adds that focusing on three to five core KPIs is usually better than showing twenty line charts (Eleken, “Why users ignore dashboards”).
How do filters and drill-down work?
Filters help when the dimensions are clear: period, location, product, or segment. The problem appears when they stack up. Girard warns that excessive filters and drill-down paths create analysis paralysis instead of speeding up decisions: a dashboard with many filters can answer almost any question, but most people won’t know where to start.
Drill-down, for its part, is the pattern of showing the summary first and the detail on request. Girard describes clicking a summary metric to see its breakdown, and advises relegating detail to drill-down pages. It is useful, but it doesn’t replace prioritization: if a metric doesn’t matter in the main view, sending it to a drill-down doesn’t make it more important.
How do you separate executives and analysts?
Girard distinguishes executive dashboards, with KPIs aligned to strategic objectives such as revenue growth or pipeline velocity, from channel dashboards, with conversion rate or cost per acquisition. Eleken proposes splitting a “master” dashboard into role-based views, because each group (executives, marketing, product, operations) should see only what is relevant to its work.
In practice, executives need to see at a glance whether things are going well or badly. Analysts need to drill into detail to understand why a figure changed. Those are different needs, and one design doesn’t serve both well.
Example
Illustrative example: a logistics platform team has a dashboard with twenty-one charts and nine filters. Nobody uses it to decide. The team rebuilds it as two views. The first has four KPI cards: on-time deliveries, cost per shipment, open incidents, and satisfaction. Below them, a weekly trend of on-time deliveries and a breakdown by route that opens on click. The second view, for operations, has detail by driver and filters for shift and zone. A month later, the weekly meeting starts from the dashboard instead of a spreadsheet.
When NOT to use a dashboard
- When the question is a one-off. If someone needs a figure once, a report or a query is enough.
- When nobody will act on it. A dashboard without linked decisions gets ignored within weeks.
- When the data isn’t reliable. A dashboard with doubtful figures breeds distrust, which is harder to fix than any chart.
- When the metric is a single number. A lone card doesn’t need a full dashboard.
Common mistakes
- Adding a metric for every area that asked for one.
- Using decorative charts or colors that mean nothing, or using color to show quantities.
- Hiding the main figure at the bottom of the screen.
- Stacking filters instead of choosing two or three key dimensions.
- Not stating which action a change in a metric should trigger.
How to start tomorrow
Open the dashboard your team uses most and write down the main question it answers. Mark the three metrics that really answer it. For each one, write what someone would do if it changed color. If there is no action, remove it from the main view and move it to a breakdown. Then review the filters and keep only the ones people actually use. If the analysis goes beyond a snapshot, funnel analysis gives you a way to read a flow step by step.