Charts and data visualisations are often the most important part of a research document. They carry findings, support arguments and shape how evidence is understood. They’re also, consistently, the most common accessibility failure in research output.

The problem isn’t a lack of care. It’s that most data visualisation is designed to look right, not to work for everyone. And the decisions that make a chart look polished – brand colours, clean minimalism, subtle contrast – are often exactly the decisions that make it inaccessible.

This article explains what accessible data visualisation requires, where the most common failures occur, and why the problem usually starts with brand colour decisions made long before any chart is drawn.

The most frequent accessibility failure in data visualisation is colour-only communication. A chart uses colour to distinguish between data series, categories or values, with no other visual encoding. For anyone who is colour-blind, has low vision, or is reading the document in greyscale, the chart becomes ambiguous or meaningless.

Colour blindness affects approximately 8% of men and 0.5% of women with Northern European ancestry. In a research audience of any size, a chart that relies solely on colour to communicate is excluding a significant proportion of readers.

But colour-only communication is only one issue. Accessibility failures in charts also include insufficient contrast between data elements and backgrounds, text labels that are too small or too light, chart titles and axis labels that are visually present but not accessible to screen readers, and charts exported as flat images without any alt text or data description.

WCAG 2.2 contrast requirements apply to text within charts just as they do to body copy. Data labels, axis labels, legend text and annotations all need to meet the 4.5:1 contrast ratio for normal-sized text. Large text labels (18pt+ or 14pt+ bold) need 3:1.

Where it gets complicated is with the data itself. WCAG 2.2 Level AA doesn’t specify a minimum contrast ratio for non-text graphical elements – it requires 3:1 contrast between the graphical element and adjacent colours. That means a bar in a bar chart needs 3:1 contrast against the background. Two bars sitting next to each other need to be distinguishable without relying solely on colour.

In practice, most research brand palettes are not tested against these requirements. They’re tested against each other visually, on screens, at full saturation. When those colours appear at smaller sizes, on different backgrounds, in print or in projected slides, the contrast often falls short.

Here is where the problem becomes structural. In most research organisations, the brand palette is fixed. Colours are specified in brand guidelines, built into templates and applied consistently across all output. When that palette fails accessibility contrast requirements for data visualisation, every chart produced using it is a compliance risk.

This is exactly the kind of upstream decision that creates downstream problems. The brand colour was chosen for its visual appeal or its resonance with the organisation’s identity. No one tested it in a bar chart against a white background at 8pt label size. No one checked whether it was distinguishable from the secondary colour for someone with deuteranopia.

The solution isn’t to abandon brand colours. It’s to develop a data visualisation palette that works within the brand but is specifically designed and tested for accessibility. We cover how this connects to wider brand system design in Article 6.

Accessible data visualisation uses multiple encodings, not just colour. That means combining colour with pattern, shape, texture or direct labelling so that the information can be understood regardless of how colour is perceived.

In practice, this often means using direct data labels rather than relying on a legend, using patterns or hatching in addition to colour fills for area charts, using distinct shapes for scatter plot data series, and choosing colour palettes that are distinguishable under common forms of colour blindness.

Tools like Coblis simulate how charts appear under different types of colour vision deficiency. Testing charts before publication is straightforward. The barrier is usually that it’s not built into the process.

When a chart is published in a PDF or on a webpage, it needs alt text that communicates what the chart shows – not just what type of chart it is.

“Bar chart showing annual research outputs 2019–2023” is not sufficient. “Bar chart showing annual research outputs 2019–2023. Outputs increased from 42 in 2019 to 87 in 2023, with the largest year-on-year increase between 2021 and 2022” is closer to what’s needed.

The goal of alt text for data visualisation is to ensure the information contained in the chart is available to someone who cannot see the chart. That requires actually describing the data, not just the visual format.

For complex charts with large datasets, consider providing a data table alongside the chart, or linking to the underlying dataset. This serves both accessibility and research transparency.

Accessible charts aren’t a compromise on quality. In most cases, the changes that make charts accessible – higher contrast, cleaner labelling, multiple encodings, descriptive alt text – also make them more effective for everyone.

Charts that work without colour, that have clear direct labels, that communicate the key finding without requiring careful interpretation – these are better charts by any standard. Accessibility and clarity point in the same direction.

If your current report templates produce charts that fail accessibility checks, the fix starts upstream: with the colour palette, the chart templates and the export process. Get in touch to discuss a report design project.