Alt Text

Alt text is the value of an img element’s alt attribute: the words that stand in for the image when it cannot be seen, whether because the reader is blind, the connection dropped it, or the client fetching the page does not render images at all.

What the HTML standard asks of it

The requirement is laid out in the HTML standard’s section on images. The text should serve the same purpose the image serves. For an image that is a link or a button, that means saying what the link does. For a chart, a map or a photograph that carries information, it means stating what the image communicates. For a logo or an icon, a short phrase naming the thing. A purely decorative image gets an empty attribute, alt="", so that a reader who cannot see it is not told about it. The standard also says what alt text should not be: a copy of a caption already on the page, or a file name.

Who reads it besides a screen reader

Any client that does not render images, which includes every text-only crawler. Google’s image documentation says alt text is used together with computer vision and the surrounding page to work out what an image is about, and warns that stuffing the attribute with keywords reads as spam. An assistant’s retrieval fetches HTML and reads the text; a photograph of a shopfront with the address painted on it contributes nothing unless the alt text or the page says the address. This site’s audit crawler ran without JavaScript and without images, which is the condition alt text is written for. It did not count alt attributes, so there is no first-party figure here.

A common question

Does alt text matter for an AI assistant if no screen reader user visits the page?

Yes, because a screen reader is not the only client that never sees the image: every text-only crawler reads a page without rendering pictures, and an assistant's retrieval works from the HTML it fetched. A photograph of a shopfront with the address painted across it contributes nothing to that reading unless the alt attribute or the page itself states the address. Google treats the attribute as one input into working out what an image is about, while warning that packing it with keywords reads as spam. No first-party figure exists, since the audit crawler did not count alt attributes.

Sources

Requirements: the HTML standard’s section on images. Search use and the warning against keyword stuffing: Google’s image documentation. The audit method (no JavaScript, no images) is described on the methodology page.