Introduction
A dashboard is a screen that people come back to, every morning or every week, to see how things are going and decide what to do. Good dashboards answer a few questions at a glance; bad ones show every chart someone could think of. Designing one brings together everything in this course: choosing charts, scales, colour, annotation and aggregation.
How it works
Start with decisions. Ask who will use the dashboard and what they decide with it. A sales manager needs revenue against target and the pipeline; a support lead needs open tickets and response times. Pick a handful of KPIs, usually three to seven, that inform those decisions. Everything else belongs in a separate report.
KPI cards. A number alone means little. A KPI card shows the value, formatted for reading (12,500 rather than 12500), its change from the previous period ("up 8.7%"), and its status against a target, for example on track at 100% or more, at risk from 90%, and off track below. A small sparkline can show the recent trend. Status uses colour plus a word or icon, never colour alone.
Layout. Readers scan from the top left. Put the most important KPIs in the first row, then the charts that explain them, then details such as tables. Most tools use a 12-column grid: a card might take 3 columns (four in a row), a main chart 8 and a side chart 4. Panels fill a row from left to right and wrap to a new row when the next one does not fit. Align panels and keep spacing even.
Consistency. Use the same colour for the same thing everywhere (for example, this year in a strong colour and last year in grey), the same date range across panels unless labelled otherwise, and shared scales for charts meant to be compared. Filters, such as region or period, should apply to every panel and show their current setting.
Trust. Show when the data was last updated, define each metric in a tooltip or footnote, and make the dashboard fast; a dashboard that takes a minute to load is not checked. When a number looks wrong, users stop trusting the whole page, so test the calculations against a known report.
Example
# KPI cards: the value, the change against last period and the status against a target.
kpis = [("Revenue", 12500, 11500, 12000), ("Signups", 90, 120, 100), ("Orders", 1000, 1000, 1100)]
for name, value, previous, target in kpis:
change = (value - previous) / previous * 100
arrow = "up" if change > 0 else "down" if change < 0 else "flat"
pct = value / target * 100
status = "on track" if pct >= 100 else "at risk" if pct >= 90 else "off track"
print(f"{name:8} {value:>7,} {arrow} {abs(change):4.1f}% {pct:5.1f}% of target, {status}")Line by line
- Each KPI has a current value, the previous period's value and a target.
- The change compares with the previous period and gets a direction word.
- The status compares with the target using the 100% and 90% bands.
:>7,right-aligns the value with thousands separators, so the cards line up.
Expected result
Revenue 12,500 up 8.7% 104.2% of target, on track
Signups 90 down 25.0% 90.0% of target, at risk
Orders 1,000 flat 0.0% 90.9% of target, at risk
In the real world
Companies run on dashboards built in tools such as Power BI, Tableau, Looker and Grafana, and in custom web pages. Operations teams watch live system dashboards, and executives review weekly KPI pages. The same principles apply everywhere: few metrics, context for every number, clear layout.
Common mistakes
- Showing too many metrics, so nothing stands out.
- Numbers without context: no comparison, no target, no period.
- Inconsistent colours and date ranges between panels.
- No update time, so users cannot tell whether the data is current.
Best practices
- Write down the audience and their decisions before choosing any chart.
- Give every KPI a comparison and, where possible, a target.
- Use a grid, put the most important panels top left, and show the update time.
Summary
A dashboard shows a few KPIs that inform its audience's decisions. KPI cards give each value with its change and status against a target, panels are laid out on a grid by importance, colours and scales are consistent, and update times and definitions keep it trustworthy.
Quick check
Key takeaways
- Start from the decisions the audience makes and show only the metrics that inform them.
- Every number needs context: a comparison period, a target or both.
- Put the most important information top left, keep scales and colours consistent, and show when data was updated.
Practice problems for this lesson
My notes
Write down what you want to remember from this lesson. Notes are saved on this device as you type, and appear on your dashboard.