Professional Development
What to remove from a confusing data chart
Decluttering a chart is a sequence of decisions about its purpose, labels and context. Remove visual effort while preserving the information readers need to judge the result.
A crowded chart can make a straightforward question difficult to answer. Eight colored lines compete with a shaded background, every point carries a label and the legend sits far from the data. The first useful edit is to decide which comparison the reader needs to make.
Removing visual elements can help, but removal is not automatically an improvement. A cleaner chart can become misleading if it loses the baseline, uncertainty or explanation that makes its numbers interpretable. Edit for a reader’s task and keep a record of what the edit changes.
Name the comparison before changing the design
Suppose a service manager wants to know whether waiting times have changed across teams. The chart also contains request volumes, a target line and two years of weekly labels. These may all be useful data, but they do not necessarily belong in one view.
Write the question above your working copy: how did waiting times change for each team over this period? Identify the unit, time range and population being compared. If the real question concerns whether a target was met, keep that target visible and reconsider which other information belongs beside it.
The UK Government Analysis Function’s chart guidance starts with the message and statistical relationship, and recommends considering whether a chart is trying to communicate too much. That is a useful diagnostic before making cosmetic adjustments.
Remove competing decoration first
Try removing background pictures, shadows, heavy borders and effects that add apparent depth. Reduce the visual weight of gridlines so they provide reference without dominating the data. Check the result at the size people will actually see it, especially if it will be embedded in a report or viewed on a phone.
Do not turn minimalism into a fixed rule about every illustration. Parsons and Shukla’s 2020 research manuscript reports interviews with twenty data-visualization practitioners and finds varied understandings of chartjunk and its uses. This is evidence about practitioner perspectives, not proof that one visual style produces better decisions in every setting.
Ask of each element: does it identify data, support a comparison, explain context or help the intended reader? If its role is unclear, remove it temporarily and see whether anything necessary becomes harder to understand.
Replace unnecessary matching and repetition
A legend makes readers look back and forth between names and colored marks. For a line chart, try naming each series beside its line where there is room. Microsoft’s PowerPoint accessibility guidance demonstrates this change when a chart becomes difficult to interpret without color. Direct labels can still collide, so inspect their placement.
Delete repetitive point labels if the task is to understand a trend rather than read every value. Retain labels for a few genuinely important points, with the full data available separately. Reduce date labels while keeping the intervals clear; a reader should still know whether adjacent marks represent weeks, months or irregular observations.
If several series remain difficult to distinguish, consider separate small charts using consistent scales. Another option is to highlight a relevant series while keeping comparison series visible in a quieter style. Explain the selection so the design does not imply that an inconvenient result was absent.
Protect the information that qualifies the story
Before removing anything further, check the details that make the chart honest:
- Scale: Does the axis still represent magnitude appropriately? Bar lengths normally need a zero baseline.
- Scope: Can the reader identify the measure, population, period and any excluded categories?
- Uncertainty: Have intervals, missing observations or provisional values been explained rather than silently erased?
- Comparison: Have you kept the relevant reference, target or earlier period?
- Provenance: Can someone find the specific source and underlying data?
Move lengthy supporting detail into nearby notes or a companion table when that preserves access and meaning. Do not delete an outlier or shorten a time range merely because it weakens the intended message. Investigate whether the observation is valid and explain any justified exclusion.
Show the revised chart to someone less familiar with the analysis. Ask them to state its main comparison and identify one limitation. If their answer differs from what the chart supports, revise again. The successful edit leaves the reader with less decoding work and enough evidence to question the conclusion.

