Designing a KPI Dashboard Executives Can Read in 30 Seconds
Design principles for executive KPI dashboards: hierarchy, context for every number, colour, and using AI to critique your dashboard before it ships.
An executive opens the dashboard on a phone while waiting for a meeting, scrolls through two screens of six pie charts and a 40-row table, then closes it with the same question: "So, are we doing well this month or not?" If you have built a dashboard and received that question back, this article covers design principles that help readers grasp the big picture quickly, and how to use AI to review your work before you send it.
What is an executive KPI dashboard?
A KPI dashboard is a summary page of key indicators showing business status against goals. Unlike an analytical report where people dig for causes, an executive dashboard has one job: answer "Where are we now, and what needs attention?" Everything else belongs on the next page.
Principle 1: Clear hierarchy
The most important thing goes top left
Readers of Thai and English scan left to right, top to bottom. Put the 3 to 5 most important KPIs in the top row, followed by charts that explain trends, then detail tables.
Limit the count
If everything is important, nothing is. Try a rule: no more than 5 KPIs and 4 charts on the first page. Move the rest to drill-down pages.
Principle 2: Every number needs context
"Sales of 12.4 million baht" says nothing until you know the target and last month's figure. A good KPI card has:
- The current value, rounded for reading, such as millions of baht instead of an 8-digit number
- A comparison to target or to the same period last year
- A small sparkline showing the last 8 to 12 periods
Principle 3: Use colour to direct attention, not to decorate
Use a neutral colour such as grey as the base and a strong colour only where something is unusual. If every chart is brightly coloured, executives will not know where to look. Also avoid relying only on red and green, because people with colour blindness cannot tell them apart. Add arrows or plus and minus signs.
Principle 4: Chart titles are conclusions
Instead of "Sales by region", write "North below target for 2 months in a row". Readers get the answer before reading the chart, but the title must be updated to match real data each time.
Principle 5: Design for mobile screens
Many executives open dashboards on their phones while on the move. If your tool supports a separate mobile layout, build one, stacking KPIs vertically and dropping long tables. If it does not, at least check that key numbers are readable without zooming and that no chart requires horizontal scrolling.
Example: a parcel delivery company dashboard
Imagine a Thai delivery company with distribution centres in several regions. The top row has 4 KPIs: parcels delivered, on-time delivery rate, cost per parcel and number of complaints, each compared to target. The middle row is a line chart of weekly on-time rate by region. The bottom row lists centres below the threshold. Within 30 seconds the executive sees where the problem is, then clicks through for detail.
If an executive has to ask "Is this number good or bad?", the dashboard is not finished.
Using AI to design and critique
- Screenshot the dashboard and ask an AI that can read images: "If you were a CEO with 30 seconds, what would you understand and where would you be confused?"
- Have AI write conclusion-style chart titles from numbers you provide, then check them yourself.
- Have AI help prune KPIs by asking: "Which of these are outcomes and which are drivers?"
Watch out for
- AI tends to suggest adding charts, rarely removing them. You have to do the cutting.
- AI may misread values from a screenshot. Do not let AI summarise numbers from images; use the real data.
- Red, amber and green thresholds must come from agreed targets, not numbers AI made up.
- Avoid 3D pie charts and bar charts whose axis does not start at zero, because they distort the size of differences.
What to do next
- Ask executives what 3 questions they want answered every morning.
- Match each question to one KPI with a target and a comparison.
- Sketch the layout on paper before opening any tool.
- Build a first version, then have AI critique it from an executive's point of view.
- Test with real users and time whether they understand it within 30 seconds.
To see the full process from raw data, read Building Dashboards with AI, or practise with the course Building a complete Excel Pivot Dashboard on Plants.work Learn.