Alt text in e-commerce: why it matters and how to write it

What it is for

Alt text —an image's alt attribute— does two jobs at once:

• Accessibility. Screen readers read it aloud to people who cannot see the image. For them, a photo without alt text does not exist. The W3C accessibility guidelines require all non-text content to have a text alternative.
• Search engines. Google uses alt text, together with the page content, to understand what an image is about. A good description helps your photos show up in image search.

On top of that, several marketplaces ask for it to publish a product's images.

What to say

The W3C sums it up in one question: what information does this image provide? On a product page, that is usually:

• which product it is, without repeating the full name and code;
• what this particular photo shows: the front, the back, a detail, the product in use;
• what is not in the text: the exact color of this variant, the size compared with a hand, what is in the box.

An example: instead of “drill”, “cordless hammer drill seen from the side, with the battery attached and the chuck open”.

What to avoid

• Keyword stuffing. “Drill cheap drill cordless drill” describes nothing and Google treats it as filler.
• Starting with “image of” or “photo of”. The screen reader already says it is an image.
• The same text on every photo. Each photo exists because it shows something different; the text should say what.
• Making things up. If the photo does not show outdoor use, the text should not suggest it.
• Leaving it empty on a product photo. Empty alt text tells the screen reader the image is decorative.

Alt text in every language

If you sell in several languages, alt text gets translated too. An English product page with images described in Spanish confuses buyers who use a screen reader and the search engine of the other market. Each image should have its description in every catalog language, and the store should use the one for the page's language.

The scale problem

Writing good alt text for one photo takes a minute. For ten thousand photos it is more than a hundred hours, which is why many catalogs never do it or fill it with the product name. It makes sense to rely on AI that looks at the image and proposes the description, as long as someone checks that it describes what is actually there.

How to review what AI proposes

When AI proposes the text, the review is quick if you look at the right things:

• Does it describe this photo or the product in general? If all ten photos say the same thing, what makes each one different is missing.
• Does it say something that is not visible? A feature, a use or a material the photo does not show does not belong in the alt text.
• Do the color and the variant match the product record the image is linked to?
• Is it in the right language for each version of the record?

With those four questions, reviewing a hundred images takes minutes, not hours.

How Bruxel handles it

In Bruxel each image lives in your library, linked to one or more products, and the “Write with AI” button generates its alt text in four languages at once: Spanish, English, French and Portuguese. The product record shows the text in the screen's language and the text travels with the image on export. The images guide explains it, and the photos for the whole catalog use case shows it on video.

Where to start

Start with the main images of your best sellers: they are the most viewed and the ones search engines show first. Check that each one has a text saying what the product is and what is shown, then move on to the gallery photos.