What you need to know
Editorial data visualisation sits at the intersection of journalistic accuracy and design legibility. A chart published in a UK newspaper or news website must communicate a finding clearly to a general reader within seconds — unlike an analytical dashboard built for specialists who can spend time interpreting complex graphics. The primary design question is always: what is the one thing this chart needs to show? Everything else — colour, annotation, secondary data series — should serve that purpose or be removed.
The Office for National Statistics (ONS) publishes a data visualisation style guide used as a benchmark by many UK public-interest publishers. It covers chart selection, colour palettes, typography, and accessibility. The Datawrapper Academy and Flourish documentation provide tool-specific guidance aligned with newsroom workflows. Between these resources, UK journalists have access to well-tested, peer-reviewed standards that do not require a design degree to apply.
Accuracy is non-negotiable. A visually attractive chart that misrepresents data — through a truncated y-axis, cherry-picked date range, or misleading percentage base — is a journalism failure, not just a design error. Always start from the underlying dataset, clean it, and apply editorial judgement about what the data actually shows before choosing how to display it.
When data visualisation issues arise
- 1Choosing a chart type before fully understanding the data — for example, reaching for a pie chart before checking whether the segments actually sum to a meaningful whole.
- 2Using a truncated y-axis that starts above zero on a bar chart, making small differences appear dramatic — this is a common source of reader complaints and corrections.
- 3Publishing a chart with no source line, making it impossible for readers to verify the underlying data or for corrections to be traced.
- 4Applying a sequential colour palette (e.g. light-to-dark blue) to categorical data where no ordering exists, implying a ranking that is not in the data.
- 5Overloading a single chart with multiple data series, annotations, and a legend when two separate charts would be clearer.
- 6Using a map to show data that is better shown as a ranked bar chart — maps emphasise geographic area, not data magnitude, and can mislead when large rural areas dominate visually.
- 7Failing to account for population differences when comparing raw counts across regions — crime figures, for instance, should almost always be shown per 100,000 population.
- 8Exporting a chart at screen resolution rather than print resolution, resulting in blurred graphics in print editions.
Red flags
- A bar chart whose y-axis does not start at zero — this is almost always misleading for bar charts (though not necessarily for line charts).
- A 3D chart — three-dimensional rendering distorts the perceived size of segments and bars and provides no additional information value.
- A pie chart with more than five segments, or where the smallest segments are unlabelled — consider a ranked bar chart instead.
- A chart where the headline or title states a conclusion that the data shown does not support — the chart must independently demonstrate the claim.
- Data sourced from a press release or lobby group without cross-referencing against an official or independent dataset.
- A colour scheme that relies solely on hue to distinguish data series with no pattern, shape, or direct label as a fallback for colour-blind readers.
Data visualisation checklist
- I have identified the single key finding this chart needs to communicate and written a headline that states it plainly.
- I have selected a chart type appropriate to the data structure: bars for categorical comparison, lines for time-series, scatter for correlation.
- The y-axis on any bar chart starts at zero.
- Every chart carries a source line naming the originating organisation, specific dataset, and release date.
- I have checked the colour palette using a colour-blind simulator (available in Datawrapper and via browser extensions).
- All data labels, axis labels, and legends are legible at the intended display size without zooming.
- If the chart shows rates or proportions across geographic areas, I have verified the population base used for normalisation.
- I have confirmed the data with the originating source or a second independent dataset before publication.
Data journalism tools & guides
Explore our Data Journalism hub for spreadsheet skills, FOIA data requests, and investigative data techniques — and our Multimedia Skills guide for combining charts with video and photo storytelling.
Common mistakes
- Treating the chart as decoration rather than evidence — if a chart does not add information beyond the accompanying text, it should not be published.
- Downloading a chart template and publishing it without updating the source line from the template's example data.
- Using Flourish's animated race chart for data that changes too slowly to benefit from animation, adding distraction without editorial value.
- Conflating correlation and causation in a scatter chart headline — a scatter plot shows association, not causal direction.
- Publishing a chart built on provisional data without flagging that the figures are subject to revision.
- Choosing a diverging colour palette (red-to-blue) for data that is not naturally bipolar, implying a political or value-laden framing that may not be intended.
Related guides
Primary sources
- Datawrapper Academy— Datawrapper
- Flourish — Data Visualisation & Storytelling— Flourish
- ONS Visualisations— Office for National Statistics
- RAWGraphs — Open Visualisation Framework— RAWGraphs
- Observable — JavaScript Notebooks— Observable
- Office for Statistics Regulation— UK Statistics Authority
Frequently asked questions
When should I use a bar chart rather than a line chart?
Use a bar chart to compare discrete, unrelated categories — for example, crime rates across different police force areas or party vote shares at a single election. Use a line chart to show change over time for continuous data, such as monthly unemployment figures or weekly COVID case counts. A common mistake is using a line chart to connect unrelated categories, which implies a trend that does not exist in the data. If your x-axis is time and your data is continuous, default to a line chart; if your x-axis is a set of distinct groups, default to a bar chart.
How should I cite data sources inside a graphic?
Every data graphic published by a UK newsroom should carry a source line, typically positioned below the chart title and above the x-axis or legend. The source line should name the originating organisation (e.g. Office for National Statistics), the specific dataset or publication, and the release date. Where data has been cleaned, aggregated, or otherwise processed by your newsroom, add a note line stating 'Analysis: [Publication name]' alongside the original source. The ONS publishes its own chart templates that demonstrate this standard format.
What is the difference between Datawrapper and Flourish?
Datawrapper is a browser-based chart tool built specifically for newsrooms. It produces clean, responsive SVG charts and maps with a focus on accessibility and speed. Its free tier is generous and it is widely used across UK regional and national titles. Flourish is a more feature-rich platform offering animated scrollytelling, race charts, and complex interactive templates. It is better suited to longer-form digital features where interaction adds editorial value. For quick turnaround charts on deadline, Datawrapper is typically faster; for flagship data projects with bespoke layouts, Flourish offers more flexibility.
How do I make a data visualisation accessible to readers with colour vision deficiency?
Avoid using red and green together as the sole means of distinguishing data series, as this is the most common form of colour vision deficiency. Use a colour-blind-safe palette — tools such as ColorBrewer provide tested palettes for sequential, diverging, and qualitative data. Always supplement colour with direct labels, patterns, or text annotations so the chart remains readable in greyscale. Datawrapper has built-in colour-blind simulation; Flourish allows custom colour palettes. The ONS Design System also publishes colour guidance aligned with WCAG 2.1 accessibility standards.
Can I use Observable or RAWGraphs if I am not a programmer?
RAWGraphs is designed for journalists and designers without coding experience. It accepts CSV or TSV data pasted directly into the browser and produces SVG outputs that can be further refined in Illustrator or Inkscape. Observable is a JavaScript notebook environment and does require basic programming knowledge to use effectively, though it has a large library of public notebooks that can be forked and adapted. For most UK newsroom journalists without a data science background, Datawrapper or Flourish will cover the majority of everyday visualisation needs; RAWGraphs is useful for more unusual chart types not available in those tools.