Start with questions, not widgets
Many dashboards are a pile of every available metric. They look busy and answer little. Instead, interview users and write down the questions they ask when they log in: Is anything broken? Are we on track? What changed since yesterday?
Each question becomes a block. If a metric does not answer a real question, it does not belong on the first screen.
Build a clear hierarchy
- Top: two to four headline numbers with change versus the last period.
- Middle: the trends and breakdowns that explain those numbers.
- Bottom or drill-down: tables and detail for users who want to dig.
NN/g's complex-application heuristics stress keeping users oriented and reducing the memory load. A hierarchy like this does both.
Match the chart to the question
| Question | Use | Avoid |
|---|---|---|
| How is it changing over time? | Line chart | Pie chart |
| Which is biggest? | Sorted bar chart | 3D charts |
| What share is each part? | Stacked bar, or a pie with few slices | More than five slices |
| How is it distributed? | Histogram or box plot | Averages alone |
| What is the exact value? | A table or a big number | A chart with no labels |
Make it usable at a glance
- Label axes and units. Never make users guess.
- Use color to carry meaning, such as status, and keep it consistent.
- Show comparison: versus last week, target, or average.
- Let users filter by date range and segment, and remember their choice.
Do not forget the first visit
A new account has no data. An empty dashboard of blank charts feels broken. Show sample data, a short explanation of what will appear, and the one action that will start filling it.



