The Office for National Statistics is not a design studio. Nobody working there is chasing a Awwwards nomination or obsessing over whether their typeface feels “premium”. And yet, if you spend time actually reading through ONS data visualisations, you start to notice something genuinely impressive: these charts communicate extraordinarily well. They are sober, functional, and designed with a discipline that most commercial dashboard teams never achieve. That is worth pulling apart.
This is not a fan letter. There are real weaknesses in the ONS approach, and I’ll get to them. But the core methodology behind their data visualisation design is a masterclass in restraint, and restraint is the hardest thing to teach a designer who has spent three years in Figma making things look beautiful.
Why the ONS chart style is so readable (even when the data is horrible)
The first thing you notice looking at ONS charts is the absence of noise. No gradients. No 3D extrusions. No decorative gridlines. The chart area is clean to the point of feeling almost spartan, and that is not an accident. Their house style explicitly prioritises clarity over aesthetics, which sounds obvious but is remarkably rare in practice.
Their bar charts use a restrained palette: typically one primary colour for the main data series, with muted secondary tones for comparison series. The ONS colour set leans heavily on accessible combinations. They use a mid-blue as a workhorse colour, which sits well on both white and off-white backgrounds, and has solid contrast ratios for users with colour-vision deficiency. A lot of commercial dashboards still use red-green pairings for comparison metrics. The ONS almost never does this, because a meaningful percentage of the population cannot distinguish those colours reliably.
The typography choice is equally deliberate. Their publications use a clean sans-serif across chart labels, axis annotations, and source attributions. The hierarchy is strict: title at the top, subtitle directly beneath it, axis labels smaller and lighter, source line at the bottom in a noticeably reduced size. There is never any ambiguity about what you are supposed to read first. That reading order matters enormously when the data is genuinely complex, like age-stratified mortality statistics or regional employment breakdowns.
The colour palette decisions that data visualisation design uk ons charts gets right
Colour in data visualisation is one of those areas where designers consistently over-engineer things. The temptation is to build a ten-colour categorical palette because it feels comprehensive. The ONS approach keeps categorical colours to a sensible minimum, typically no more than six distinct values on a single chart, and when the data requires more categories than that, they restructure the chart rather than add more colours.
This is actually the correct answer, and most teams reach it too late, after building a twelve-colour legend that nobody can parse at a glance.
Their sequential palettes for choropleth maps (the regional breakdown maps you see for things like median household income or broadband coverage) use single-hue progressions, typically moving from a pale tint to a saturated anchor. This is textbook perceptual uniformity. The human visual system processes luminance gradients more reliably than hue shifts, so a light-to-dark single colour reads as a continuous scale far more intuitively than a rainbow palette. The ONS gets this right by default. A huge number of Local Authority and NHS data dashboards still use rainbow gradients in 2026, which is genuinely baffling.
Where the ONS approach has genuine gaps
Right. Enough praise. There are real criticisms to make.
Interactivity is sparse. The static chart approach works brilliantly for published reports and press releases, but the ONS digital presence has been slow to adopt genuinely exploratory visualisation. Their datasets are enormous and often the most interesting insights live in the sub-groups: age cohorts, regional splits, occupational categories. A well-built interactive chart could surface those without requiring a user to download a 40MB Excel spreadsheet. Tools like Observable Plot or D3.js could handle this elegantly. Some ONS pages now include simple chart builders, but the experience still feels like an afterthought compared to, say, what the Financial Times graphics team produces.
Responsiveness is another weak point. Many ONS chart embeds were designed for desktop screens and degrade awkwardly on mobile. Given that mobile accounts for a substantial share of web traffic across the board, a chart whose axis labels overlap at 375px width is simply not finished. This is a common problem across government digital estates, but it matters more for the ONS because their data is genuinely in the public interest.
And annotation. The ONS occasionally adds annotation to charts for major events (a shaded region for the pandemic period, a labelled inflection point for a policy change), but this is inconsistent. Annotation is arguably the most powerful tool in data storytelling. Telling the reader why a spike exists turns a confusing chart into a coherent argument.
What to actually steal for your own UK data dashboard
If you are building a dashboard for a UK public sector client, a fintech, or any product that has to make complex data legible to non-specialists, here is what I would take directly from the ONS playbook.
First: commit to a maximum of five categorical colours. Pick them based on contrast ratios and colour-vision accessibility, not because they match your brand guidelines. Your brand team will survive.
Second: use a strict typographic hierarchy with no more than three size levels across a chart. Title, axis labels, and source attribution. If you need a fourth level, the chart is probably too complicated.
Third: strip the gridlines back to horizontal-only for bar and line charts. Vertical gridlines rarely add information and almost always add visual weight. The ONS rarely uses them, and their charts are better for it.
Fourth: when your data has a natural comparison period (pre-pandemic vs post-pandemic, pre-Brexit vs post-Brexit trade figures), use a subtle background band to mark that period rather than relying on the user to cross-reference dates. The ONS does this consistently and it genuinely aids comprehension.
Fifth, and this is the big one: design for the worst-case reader, not the best-case reader. The ONS writes for a journalist filing a story at speed, a policy analyst who has fifty tabs open, and a curious member of the public who has never read a statistics bulletin before. All three need to extract the key finding within about ten seconds. If your dashboard only works for someone who already understands the domain, it is not doing its job.
The practical upshot for designers working with data
There is a quiet revolution happening in UK product teams around data literacy. More designers are expected to understand chart types, know when a line chart is wrong for the data they have, or recognise that a pie chart with seven segments is functionally useless. The ONS charts, for all their limitations, are a free masterclass in applied data visualisation design. They are public documents, they cover every chart type in common use, and they have been iterated over decades with a clear mandate to communicate reliably rather than impress.
Go look at them properly. Not to screenshot something pretty, but to study why something works. Then bring that rigour back to whatever dashboard you are building, and watch how much cleaner it gets when you stop trying to make it look clever.
Frequently Asked Questions
What chart types does the ONS use most often in their data visualisations?
The ONS relies heavily on line charts for time-series data, bar charts for categorical comparisons, and choropleth maps for regional breakdowns. They tend to avoid pie charts and 3D charts, which is consistent with established data visualisation best practice.
How do I make my data dashboard accessible to colourblind users?
Use single-hue sequential palettes for continuous data and limit categorical palettes to colours that differ in both hue and luminance. Tools like the Colour Contrast Analyser (free from the Paciello Group) can check your combinations. Avoid red-green pairings entirely.
Is the ONS data visualisation approach suitable for commercial or private sector dashboards?
The core principles, restraint in colour, strict typographic hierarchy, and clean chart areas, translate directly to commercial contexts. You may need to incorporate brand colours, but the structural decisions the ONS makes are universally applicable.
What tools do UK design teams typically use to build data dashboards?
Common choices include Tableau, Power BI, and Looker for business intelligence dashboards, and D3.js or Observable Plot for bespoke web-based visualisations. Figma is widely used for prototyping chart layouts before building in code.
How many colours should a data visualisation use?
Most data visualisation experts recommend a maximum of five to six distinct categorical colours per chart. Beyond that, the legend becomes too complex to parse quickly and the visual differentiation between series breaks down, particularly on small screens.