Dashboard Design Best Practices: Build Dashboards People Actually Use

In this guide I go through what a dashboard is for, its eight parts, and six rules from the free course Dashboards & Data-Heavy UX on DesignPath that turn a wall of numbers into a screen people decide from.

What Makes a Good Dashboard Design?

A good dashboard design helps somebody decide something in the time they have. It opens on a useful default view, puts the most important number first with the context that makes it meaningful, and lets people go from the overview to the detail when they need it. Everything else on the screen either supports that decision or gets in its way.

Think of the dashboard in a car. The driver looks at it for a second, and in that second they need to know whether to do anything: slow down, refuel, stop. Nobody puts twelve gauges in front of a driver.

1. Start From the Decision

Before any chart, write down who opens the dashboard, how often, and what they decide after looking at it. A support lead deciding where to send people this morning needs a different screen from a director reviewing the quarter. If you can't name the decision, the dashboard will end up displaying numbers instead of answering a question.

2. Check the Eight Parts

According to the course, eight parts appear on almost every dashboard ever built, and each one answers a different question. Most screens ship four of them and leave the reader to supply the rest. When you review a dashboard, list what each part answers and what nobody answers yet: what am I looking at, for which period, is this good or bad, and what changed.

3. Design the Default View First

Almost nobody changes the filters. Whatever the dashboard shows when it opens is, for most people, the entire product. Choose the period, the segment and the sorting for the most common decision, and make the current filters visible so nobody reads last month's numbers as today's.

4. Build Hierarchy Under Density

When everything on the screen is data, nothing stands out by default. Edward Tufte's data-ink ratio gives you the tool: remove everything that doesn't carry information, such as heavy gridlines, borders, shadows and decorative color, and what remains becomes the hierarchy. Then give the one number that matters most the most space.

5. Add Comparison and Context

4,182 is not information. Against last week, against the target, or against what it was before the change, it becomes something somebody can act on. Every key number on a dashboard needs at least one comparison next to it, and color should carry meaning, such as above or below target, instead of decoration.

6. Let People Drill Down

Ben Shneiderman's mantra is overview first, zoom and filter, then details on demand. The order matters, and most dashboards either skip the overview or never offer the detail. Let people click a summary number to see what's behind it, level by level, and give them a clear way back up.

How Do You Keep a Dashboard Useful?

Charts are added by request and removed by nobody, so every dashboard grows. The only reliable way to keep one useful is to find out which parts have never changed a decision. Ask the people who use it which charts they last acted on, check what they click, and move the rest to a secondary view.

That's the brief of the course practice, Redesign a Dashboard: take a real dashboard you use or maintain, find the two charts people use and redesign the screen around the decisions behind them. It also makes a strong portfolio case, because the before and after are easy to see.

Where to Start

Do you design or maintain a dashboard? Does it have more charts than anyone looks at? Could you name the decision each chart supports?

If you answered yes, yes and no, start with the free course Dashboards & Data-Heavy UX: six short lessons and a practice task. For the table views inside your dashboard, read data table design, and see where data work fits in your growth on your personalized UX design learning path.