Write long descriptions for marketing charts
Pair charts with short identifiers, structured long descriptions, exact data, caveats, and article copy that explains the intended takeaway.
Alt text is not a storage box for everything a chart contains.
A useful chart may encode values, categories, trends, exceptions, and relationships. Compressing all of that into one long alt attribute creates an exhausting sentence and still loses structure.
W3C's updated guidance on complex images recommends a two-part alternative. A short description identifies the image, while a long description provides the essential information. The page text should also point out the takeaway the chart is meant to communicate.
Marketing teams can turn that guidance into a repeatable publishing method.
Decide what the chart is doing
Before writing any description, name the chart's job.
Is it showing a trend, comparing groups, exposing a distribution, mapping a process, or explaining a relationship. The job determines which information is essential.
A chart showing newsletter signups over twelve months may exist to show seasonality. Another chart with the same values may exist to compare the period before and after a form change. The description should follow the argument, not simply list the data.
Write one takeaway sentence in the article body. If the team cannot agree on that sentence, the chart may not have a clear purpose.
Write a short identifier
The short description should tell the reader what kind of image it is and what it covers.
Line chart comparing weekly newsletter signups before and after the June form change is useful. Chart is too vague. A paragraph containing every point is too much.
When a long description appears nearby, indicate where it can be found. Keep the short text natural and avoid keyword stuffing.
Decorative styling does not belong in the short description unless it carries meaning. If color distinguishes series, name the series by label rather than relying on red and purple alone.
Build the long description in reading order
A long description usually needs context, scale, values, relationships, and takeaway.
Start with the population and period. Explain the axes or categories. Then present the values in a structure someone can navigate, often a table. Follow with the important trend, exception, or comparison.
Do not turn the description into a visual inventory. The reader needs equivalent information, not a record of every grid line.
For a simple chart, a few paragraphs may be enough. For a multi-series chart, use headings and a table so the structure remains available. W3C cautions that some programmatic associations can flatten structured content into one continuous paragraph, so choose an approach that preserves navigation when headings and tables matter.
Put the data in a table when precision matters
Charts are good at showing shape. Tables are good at exposing exact values.
If the reader may need the numbers, include an HTML table in the long description or adjacent content. Give it a caption, clear headers, units, and consistent rounding.
Do not paste a screenshot of a spreadsheet as the alternative. That creates another inaccessible image.
For very large data sets, provide the meaningful subset and a downloadable file when appropriate. Explain any filtering or aggregation so the text does not imply more precision than the chart provides.
Describe uncertainty and missing data
Marketing charts often hide measurement limits behind a smooth line.
State when data is partial, estimated, sampled, affected by tracking changes, or missing. Distinguish zero from unavailable. If a taxonomy or event definition changed, name the break in comparability.
The long description should preserve important caveats shown through error bars, shaded ranges, or annotations. Otherwise the text alternative can sound more certain than the visual.
Keep the article from repeating itself
The article body, short description, and long description have different jobs.
The body explains why the chart matters. The short description identifies it. The long description makes the information available without relying on the image.
Some repetition is necessary, but avoid copying the same paragraph three times. Link the pieces clearly so readers can choose the level they need.
Create the description from source data
Write the long description while the underlying data and analysis notes are available. Reconstructing it from a final PNG invites transcription errors.
Store the chart title, data source, query or transformation, units, date range, takeaway, short description, and long description together. If the chart changes, review every part of the package.
An AI tool can help draft prose from structured data, but a reviewer should verify values, ordering, comparisons, and caveats against the source. A fluent wrong description is worse than an incomplete draft because it looks finished.
Test with the image unavailable
Hide the image or replace it with its text alternative. Can a reader understand the main claim. Can they find the exact values when those values matter. Can they tell what changed and over which period.
Then navigate the page with a screen reader. Check the figure label, description link, headings, table headers, and return path. Avoid forcing someone through a huge description before reaching the next section.
Also inspect the chart visually. Accessible text does not excuse tiny labels, low contrast, or meaning encoded by color alone.
Add the description to chart QA
Before publication, require four approvals.
The source data and transformation are correct.
The chart communicates one clear job visually.
The short and long descriptions provide equivalent meaning.
The article explains the takeaway without overstating it.
Charts are often used to make marketing content feel authoritative. A long description tests whether that authority is real. If the team can explain the data clearly without the image, the visual is more likely to be purposeful, accurate, and useful to everyone.
