How do I write alt text for an image that contains important words?
Quick answer
When an image contains important words, include the same readable words in its text alternative rather than replacing them with a vague description. W3C advises that the text alternative for an image of text contain the same words as the image. [S1] If wording is unreadable or unavailable, don’t invent it: describe only what you can verify and flag the gap for checking.
The question, the details, the next step.
What should I put in alt text when an image itself contains important wording? I want someone who can’t see the image to get the words, not just know that there’s text there.
Include the wording that is actually readable in the image. W3C advises that when an image of text is used, its text alternative should contain the same words as the image. [S1] A practical drafting pattern is: identify the kind of item, reproduce its important readable wording, and briefly state its role if that helps make the wording understandable. Keep the description tied to what the image shows; don’t replace its words with a guess about what they might mean.
Can you show me that pattern with an example? Suppose I have a campaign graphic with the words “Library opens late Thursday” and a smaller line saying “Until 8 p.m.”
Assuming both lines are clearly readable and are the wording you want conveyed, you could draft: “Library schedule notice: ‘Library opens late Thursday.’ ‘Until 8 p.m.’” That preserves the two visible lines and labels the item without adding a reason or extra details. The label is an editorial aid, not a required phrase. Check the wording against the image; if the graphic has other important text, account for that too.
What if one line is blurred or cut off? I can make out the start, but I’m not sure about the ending.
Don’t complete the line from context. The source supports reproducing supplied, readable wording, but it does not establish wording that cannot be read or has not been supplied. [S1] Keep any text you can verify, then mark the uncertain part as a drafting gap outside the proposed alt text—for example, “[check the final word against the original image].” If an approved copy is available, use it to resolve the gap before finalizing. If not, describe only what is clear and avoid presenting a guess as exact text.
How do I keep the draft useful when the image has several bits of text, such as a heading, a date, and a short call to action? I also want an easy way to review it.
List the visible wording in a sensible order, such as heading, date, then call to action, and preserve their relationships with simple labels. For example, assuming these are the exact readable words: “Event notice. Heading: ‘Community clean-up.’ Date: ‘Saturday, May 18.’ Call to action: ‘Sign up today.’” This is a made-up example, not verified event information. Before using your own draft, compare every phrase with the image. If a phrase is unclear, leave a review note rather than silently filling it in.
Could you give me a reusable template, and a quick way to decide whether I’m done?
Try this copyable pattern: “{kind of image or notice}. {Text role}: ‘{exact readable words}.’ {Next text role}: ‘{exact readable words}.’” Use only the roles that help explain the image, and leave out placeholders you cannot verify. Then review the draft against the image: check that each important readable phrase is represented, that you have not altered or inferred wording, and that any unresolved text is flagged for checking. The template is a writing aid, not a substitute for checking the actual image or an approved text copy.
A practical checklist
Identify the image or notice briefly, if that helps orient the reader.
Copy each important readable phrase accurately and check it against the image.
Put separate phrases in a sensible order and label their roles only when useful.
Do not infer obscured, cropped, or unavailable words from context.
Mark unresolved wording for verification before treating the draft as final.
A few more useful answers
Should I summarize the wording instead of reproducing it?
For an image of text, W3C advises that the text alternative contain the same words as the image. [S1] A short label can add context, but it should not replace the readable wording.
What if I only have a description of the image, not the image or its text?
Use only wording that has actually been supplied and can be verified. The cited guidance does not establish what missing or unreadable text says, so leave a clear gap for checking rather than guessing. [S1]
Do I need to use labels such as “Heading” or “Date”?
No. They are optional drafting aids for showing how separate phrases relate. Keep a label only if it makes the wording easier to understand, and don’t add details that the image does not establish.
Sources
References used for this guide.
- [S1] Images Tutorial: Text Alternatives for Images Containing Text w3.org · Checked