Alt text for charts and graphs in PDFs: the short description, the long description, and when to use a data table

A photograph can usually be described in a sentence. A chart cannot, because a chart's job is to let a reader compare, rank, or spot a trend across many values, and no single sentence carries that. This guide sets out the pattern that works for charts, graphs, and infographics in PDFs, using the two-part alternative W3C recommends and the tagging PDF/UA requires.

The two-part pattern

W3C's Web Accessibility Initiative tutorial on complex images states that such images require "a two-part text alternative": the short description "to identify the image and, where appropriate, indicate the location of the long description", and the long description, "a textual representation of the essential information conveyed by the image". On the web the long description can be linked, grouped with the figure, or referenced with aria-describedby; the tutorial notes that the last of these only works for text-only descriptions without structure. In a PDF the equivalents are simpler because there is no hyperlink convention to lean on: the long description is simply the tagged text, list, or table that follows the figure in reading order.

Where each part lives in a PDF

• Short description: the Alt entry on the Figure structure element (WCAG technique PDF1). PDF/UA-1 requires it; Matterhorn failure condition 13-004 is "<Figure> tag alternative or replacement text missing". Acrobat's checker reports the same rule as Figures alternate text.
• Caption: a Caption element inside or adjacent to the Figure (Matterhorn 13-003 expects captions to be tagged as captions). A caption names the chart; it is not a substitute for the alternative.
• Long description: tagged paragraphs, a list, or a Table with TH header cells, placed immediately after the figure in the reading order so that a screen reader user meets it in sequence. If it lives elsewhere, the short description must say where.

Writing the short description

The short description has three jobs: say what kind of image it is, say what it is about, and give the reader the same takeaway a glance gives a sighted reader. A useful formula is type, subject, conclusion, pointer:

• "Bar chart of complaints received per month in 2025. Complaints fell from 412 in January to 168 in December. Figures are in the table that follows."
• "Line graph comparing average response time for three departments, 2021 to 2025. Housing improved most, from 14 days to 5. Full data in Table 3."
• "Pie chart of budget by programme. Education is the largest share at 41 percent; the other five programmes are each under 20 percent. Values are listed below."

Leave out colours and visual styling unless colour itself is the subject, do not begin with "image of" since the Figure tag already announces it, and do not repeat the caption word for word. Keep the alt to a few sentences; the detail belongs in the long description.

Writing the long description

For a chart with discrete values, the long description is a data table: one row per category or period, header cells for the axis labels, units in the header. Tagged correctly (a Table element with TH cells and Scope attributes, per Matterhorn checkpoint 15), it gives a screen reader user something a sighted reader does not have: the exact values, navigable cell by cell. For a chart that shows a trend or distribution rather than a handful of values, write it as prose: the range of each axis, the overall shape, the notable points (peaks, crossovers, outliers), and any comparison the chart exists to make. Canada's digital accessibility toolkit, in its guidance on alternative text and long descriptions, gives the same advice: describe scales, values, relationships, and trends, and present the description as a table where that fits.

Average response time in days, 2021 to 2025.
Housing: 14, 12, 9, 6, 5. Planning: 20, 19, 19, 17, 16.
Licensing: 8, 8, 7, 7, 7.
Housing fell every year and overtook Licensing in 2024;
Planning improved by four days over the period.

Infographics

An infographic is a layout, not an image: several charts, callouts, icons, and a story arranged for visual scanning. The accessible version is the story reconstructed as structure. Give the infographic a Figure with a short description of its purpose, then follow it with headings for each section, the callout text as paragraphs, and each chart's data as a table. If the infographic is the whole document, the document should probably not be an image at all; publishing the same content as an accessible page is the better path, and the remediation options page explains how DocAccessible produces one.

Charts that are decorative, and charts that are text

Two edge cases. A chart repeated purely for visual rhythm, such as a sparkline that duplicates a number already in the text, can be an artifact (WCAG technique PDF4) so long as no information is lost. And a chart whose values are already printed on it as real text, such as a bar chart with data labels, still needs a Figure alt, because the labels are extracted as a string of numbers with no relationship to their bars; the long description restores the relationship.

Checking your work

1. Open the PDF in the free tag viewer and confirm every chart is a Figure with Alt, followed in the tree by its description or table.
2. Run the PDF checker; it flags figures with no alternative but cannot judge the wording, which is the next step.
3. Read the alt and long description aloud without looking at the chart. If you could not explain the chart's point to a colleague from what you heard, rewrite it. The screen reader test makes this concrete.
4. Where descriptions are drafted automatically, treat them as drafts. DocAccessible's editor drafts alternative text for review and treats it as unconfirmed until a person accepts it, because software can describe what a chart shows but not what it is for.

Frequently asked questions

Sources

The statements above rest on the primary sources below. Where a source is a law or standard, the version and date named in the text are the ones checked on the review date. This guide is general information, not legal advice.

1. W3C WAI: Images Tutorial, Complex Images . The two-part text alternative and the approaches for long descriptions.
2. W3C WAI: PDF1, applying text alternatives to images with the Alt entry
3. W3C WAI: PDF4, hiding decorative images with the Artifact tag
4. PDF Association: Matterhorn Protocol 1.1 (PDF) . Failure conditions 13-003, 13-004, and checkpoint 15 on tables.
5. Government of Canada Digital Accessibility Toolkit: Alternative text and long description, best practices