10 data visualization mistakes to avoid in 2026

A chart has one job: let someone see a pattern in numbers faster than they could by reading a table. Most bad charts fail at that job in one of a small number of ways. Here are the ten we see most, with a fix for each.

1. The wrong chart for the question

The chart should follow the question, not the other way round:

Common mismatch: a line chart across unordered categories such as regions. A line implies a sequence, so readers look for a trend that does not exist. Use bars.

Fix: write the question as a sentence above the chart. If the chart does not answer that sentence at a glance, change it. Our guide to widget types lists which chart suits which question.

2. Bar charts that do not start at zero

A bar's length encodes its value. If the axis starts at 90, a bar at 95 looks twice as long as one at 92.5, when the real difference is under 3%.

When truncation is fine: line charts and dot plots, where position rather than length carries the value. A stock price moving between $140 and $150 is unreadable on an axis from zero.

Fix: bar and column charts start at zero, always. If the differences you care about are small, show the change itself (a bar chart of percentage change) rather than stretching the axis.

3. Too much on one chart

Twelve lines on one chart, labels on every point, three legends. The reader gives up.

Fix: - Show the top 5 and group the rest as "Other". - Highlight the one series that matters and grey out the others. - Split into small multiples: one small chart per series, same axes. - Move exact values to a table below the chart.

4. Meaning carried by color alone

Roughly 1 in 12 men and 1 in 200 women have some form of color vision deficiency, most often red and green. A dashboard where red means bad and green means good is ambiguous to them.

The Web Content Accessibility Guidelines (WCAG 2.1) say color should not be the only way information is conveyed (success criterion 1.4.1), and that graphical elements needed to understand a chart should have a contrast ratio of at least 3:1 against their background (1.4.11).

Fix: - Add a second cue: an arrow, a plus or minus sign, a label or a pattern. - Use a palette designed for color vision deficiency, such as blue and orange instead of red and green. - Check your dashboard with a color-blindness simulator before publishing.

5. Numbers without context

"Revenue: $184,000." Good or bad? Nobody can tell.

Fix: give every headline number a comparison: against target, against last period, or against the same period last year. Add a one-line note for anything unusual, such as "March includes a one-off $20,000 order". This matters most on executive dashboards, where readers have seconds, not minutes.

6. Related charts on different scales

Two charts side by side, revenue by region, one with an axis to $50,000 and one to $200,000. The bars look similar, the numbers are four times apart.

Fix: when charts are meant to be compared, give them the same axis range. If one series genuinely dwarfs the others, say so in the title rather than hiding it with separate scales.

7. Decoration over information

Gradients, drop shadows, background images, thick gridlines and icons on every bar all compete with the data.

Fix: remove anything that does not help the reader read a value. Light gridlines or none, direct labels instead of legends where possible, one accent color for the thing you want noticed.

8. A time range picked to tell a story

Start a revenue chart at last year's low point and any business looks like it is growing. Start it at the peak and the same business looks like it is shrinking.

Fix: - Use standard ranges (last 12 months, year to date, last 8 quarters) and use them consistently. - For seasonal businesses, compare with the same period last year. - If you do pick a custom range, say why in the subtitle.

9. 3D charts

3D pies and bars distort values: slices at the front look bigger than equal slices at the back, and bar heights are hard to read against a tilted axis.

Fix: use flat charts. There is no business chart that a 3D effect improves.

10. Dual-axis charts that suggest a link

Two lines on two different y-axes can be made to look correlated or unrelated just by changing the axis ranges. Readers also mix up which line belongs to which axis.

When they are acceptable: two closely related measures that share a time axis, such as order volume and average order value, clearly labelled and colored to match their axis.

Fix: usually better options exist: two charts stacked with a shared time axis, both series indexed to 100 at the start date, or a scatter plot if the real question is whether the two move together.

A quick check before you publish

Run through this list for every chart:

• Can I state the question it answers in one sentence?
• Does a bar chart start at zero?
• Is there a comparison for every headline number?
• Would it still make sense in greyscale?
• Are there fewer than 6 series or slices?
• Is the time range standard, or explained?
• Would a table be clearer?

Our post on presenting data to non-technical stakeholders covers the next step: turning correct charts into a clear story.

How clariBI picks charts

When clariBI answers a question or builds an AI dashboard, the chart type follows the calculation behind it: a value over time becomes a line, a top-N list a bar chart, a distribution a histogram, two numeric columns a scatter plot. That avoids mistake 1 in most cases, but not all of them. A pie chart of twelve categories is still possible if you ask for one. The chart recommendation docs explain the rules, and our deeper piece on how algorithms choose chart types covers where automatic choice goes wrong.

If you have a dashboard already, open it now and apply the checklist to the busiest chart on it. Most dashboards get noticeably clearer after removing two things and adding one comparison.