Lesson 10 of 11 · 9 min · Telling the story
Choosing the right chart and laying out a dashboard
What you will learn
- Pick a chart by the question it answers
- Avoid the charts that mislead
- Use colour, labels and titles with purpose
- Lay out a dashboard in a clear reading order
- Make dashboards accessible and phone-friendly
In simple words
A chart is a way of answering a question without making the reader do sums. The wrong chart makes them work hard or leads them to the wrong answer; the right chart shows the answer in two seconds. A dashboard is a set of charts on one page, so it needs the same care as a good poster: the most important thing first, a clear order, and nothing that distracts.
Words to know
- Chart type: the form of a chart, such as bar, line or scatter.
- Data-ink: the part of a chart that carries information, as opposed to decoration.
- Visual hierarchy: the order in which the eye reads a page, led by size, position and colour.
- Alt text: a written description of a visual for people who cannot see it.
Start with the question
Ask what the reader must understand, then choose the chart:
| Question | Chart | Notes |
|---|---|---|
| How much? (one number) | Card or KPI | show the number large, with a comparison (vs target, vs last year) |
| How do categories compare? | Bar or column | sort from largest to smallest; horizontal bars for long names |
| How does it change over time? | Line | dates on the x-axis in order; columns for a few periods |
| What are the parts of a whole? | Stacked bar, donut (2 to 5 slices), treemap | pies are poor for comparing similar slices |
| How are values distributed? | Histogram or box plot | bins matter |
| Are two measures related? | Scatter | add a trend line; size the points for a third measure |
| How did we get from start to end? | Waterfall | profit bridge, headcount change |
| Which step loses people? | Funnel | in order, with percentages |
| Where? | Map | only when location matters; a bar chart by city often reads better |
| Exact values or lookups | Table or matrix | fewer columns; conditional formatting |
Resist adding charts because they are available. A bar chart of ten sorted items beats a radar chart nearly every time.
Charts that mislead
- Truncated axis on bars. Bars encode value by length, so the axis must start at zero. (Line charts may use a non-zero axis, clearly labelled.)
- 3-D effects distort sizes; avoid them.
- Pie charts with many slices, or slices of similar size that the eye cannot rank.
- Dual axes with different scales, which can create false correlations; label them or use two separate charts.
- Too many series on a line chart (spaghetti); highlight one and grey the others.
- Unsorted bars without a reason, making ranking hard.
- Cherry-picked time ranges that hide the long-term pattern.
- Colour that encodes nothing, or red and green alone (a problem for colour-blind readers).
- Averages without context (see the statistics course): a card saying "average salary" on skewed data misleads.
Colour, labels and titles
- Use one neutral colour for most data and one accent for the point you want noticed ("Hyderabad is above target").
- Use colour consistently: the same product or region keeps the same colour across the report.
- Red and green carry meaning (bad and good) but pair them with icons, signs or text; choose a colour-blind-safe palette (blue and orange).
- Provide enough contrast between text and background (the WCAG guideline is at least 4.5 to 1 for normal text).
- Title with the message: "Mobile conversion fell 1.4 points in October" instead of "Conversion by device".
- Label directly on the chart when you can, and remove legends and gridlines that do not help. Keep fonts large enough (generally 10 pt or above for axes, bigger for titles and cards).
- Show units (₹, %, thousands) and the time period on the page.
- Add comparison to the numbers: versus target, last period, or last year, with the direction shown (arrow and percentage).
Dashboard layout
Readers scan, they do not read. Design for how eyes move: top-left to bottom-right in languages read left to right.
A reliable layout:
- Top: a title (what and for whom), the period, and the key slicers (date, region).
- Top row: 3 to 5 KPI cards with the headline numbers and their comparison. These answer "are we OK?" in five seconds.
- Middle: the main chart(s) that show the trend or the key breakdown: where the story is.
- Lower: detail (a table, a breakdown by product or city) for those who want to dig, ideally with drill-through to a detail page.
- A footer: data source, last refreshed, definitions or a link to them.
Principles:
- Limit the page: about 5 to 8 visuals. If you need more, make another page.
- Group related visuals and align them on a grid with equal spacing; use backgrounds, not borders, to group.
- Make the most important visual the largest and put it first.
- Keep a consistent design across pages: same header, slicer positions and colours.
- Use white space; clutter hides insight.
- Use navigation buttons and a clear page order for multi-page reports. A good order is overview, then breakdowns, then detail.
- Use a theme (View > Themes, or a JSON theme) so colours and fonts are consistent and easy to change.
Accessibility and mobile
- Add alt text to visuals, and give every page a meaningful title.
- Set the tab order (Selection pane) so keyboard and screen reader users meet visuals in the reading order.
- Do not rely on colour alone; use labels, patterns or icons.
- Avoid flashing and animation that distracts.
- Create a mobile layout (View > Mobile layout) with the most important cards and one or two charts stacked vertically, because many people in India open dashboards on phones.
- Make sure text is readable at phone size and the touch targets (buttons, slicers) are big enough.
Check your own work
Before you show a dashboard, ask:
- What is the one question this page answers? Is the answer visible in 5 seconds?
- Would a colleague who has not seen it explain what it shows?
- Is every axis, unit, filter and period labelled?
- Do the numbers reconcile with a trusted source?
- If I removed this visual, would anyone miss it?
A dashboard that nobody can act on is a decoration. Every visual should help a decision.
Common mistakes
- Choosing the chart before the question.
- Starting a bar chart axis above zero.
- Using 3-D or decorative pies.
- Putting twenty visuals on one page.
- Relying on red and green alone.
- Titles that say what the chart is, not what it shows.
Try it yourself
- Take a dashboard you have seen (or built) and list each visual with the question it answers; remove any that answer none.
- Rebuild a pie chart with 8 slices as a sorted bar chart and compare readability.
- Redesign a page using the five-part layout above; add a title that states the finding.
- Change a red and green chart to a colour-blind-safe palette with labels.
- Create a mobile layout with three cards and one chart.
Check yourself
Why must a bar chart's axis start at zero?
Bars show value by length, so a truncated axis exaggerates differences and misleads readers.
When is a line chart better than a bar chart?
For showing change over a continuous period of time, especially many periods, since the line shows the trend.
What goes at the top of a dashboard?
A clear title, the time period and key slicers, followed by the headline KPI cards that answer "how are we doing?" at a glance.
Why should colour not be the only way to show meaning?
Colour-blind readers and monochrome printouts lose it; add labels, icons or patterns.
Interview questions
How do you decide which chart to use?
Start from the question: comparison, trend, composition, distribution, relationship or exact lookup; then choose the simplest chart that shows the answer, avoid decoration and check readability.
What makes a dashboard effective?
A clear purpose and audience, a few meaningful KPIs with context, a logical layout with a visual hierarchy, consistent design, accurate numbers, and the ability to drill into detail.
How would you present the same data to executives and to analysts?
Executives get a short, high-level page with KPIs, trends and conclusions; analysts get detailed pages, filters, drill-through and definitions.
Summary
- Choose the chart from the question; sort bars and keep axes honest.
- Use one accent colour, consistent colours, clear labels and message titles.
- Lay out KPIs first, the main story next and detail last, on a grid with white space.
- Design for accessibility and phones, and test whether the answer is visible in five seconds.
Going deeper
- Tips for designing a great Power BI dashboard: Microsoft's design advice.
- Visualization types in Power BI: the visual gallery.
- Design reports for accessibility: alt text, tab order and colour.
- Web Content Accessibility Guidelines: contrast: the colour contrast rule.
See how much of this you can explain
A free mock interview asks you new questions on Power BI and dashboards for Data Analyst and scores every answer, with what to fix.