Missing image alt text

No `alt` attribute at all: assistive technology has no instruction. Most screen readers announce the file name, which is usually meaningless and occasionally absurd — `IMG_4471.jpg`.

`alt=""`: an explicit statement that the image carries no information. The screen reader skips it. This is the correct state for decorative images, and it is a deliberate choice, not an omission.

`alt="…"`: the text conveys what the image conveys. For a photo of a product, what the product is. For a chart, the conclusion the chart shows. For a button, the action.

WCAG 2.2 success criterion 1.1.1 (Non-text Content) is a Level A requirement: non-text content must have a text alternative serving an equivalent purpose. It is one of the most basic accessibility requirements there is.

It also matters when images fail to load — a slow connection, a blocked CDN — where the alt text is what remains.

And it gives search engines the only description they have of the image content, which is what image search uses.

Describe the function, not the appearance. An image inside a link should describe where the link goes, not what the picture looks like.

Do not start with "image of" or "picture of". Screen readers already announce that it is an image.

Keep it to what is needed. A long description belongs in the surrounding text or in a linked long description, not in an attribute.

Where the image repeats adjacent text — a thumbnail next to an article title that links to the same place — use `alt=""` so the link is not announced twice.

In the console: `[...document.images].filter(i => !i.hasAttribute('alt'))` returns the images with no attribute at all — the ones that are certainly wrong.

Then review the ones with empty alt and ask whether each is really decorative. This step cannot be automated, which is the honest limit of any tool here.

Browser developer tools include an accessibility inspector showing the computed accessible name of each element, which is what assistive technology will actually announce.

`accessibility.image_alt` reads the images of each crawled page and reports those with no `alt` attribute, with the page and the image source.

`content_image_alt_too_long` and `content_image_filename_alt` cover two adjacent faults: an alt that is a paragraph, and an alt that is the file name pasted in — technically present, practically useless.

What it does not judge is whether a given alt text is a good description, or whether an empty alt was the right call. Those need a person, and the report says so rather than scoring them.

Most images fall into one of a few shapes, and each shape has an answer that no automated check can reach on its own.

A logo in the header, linking home. The alt describes the destination, not the graphic: the company name is right, `Logo` is not, and an empty alt leaves the link with no accessible name at all.

An icon next to a text label. The text already says it, so the icon takes `alt=""` — describing it repeats the label and makes the interface twice as long to listen to.

An icon with no text label, used as a button. It carries the entire meaning, so it needs the action: `Search`, `Close`, `Add to basket`.

A chart or a diagram. The alt carries the conclusion — `Revenue grew 40% between 2023 and 2025` — and the underlying numbers belong in a table or in the surrounding text, where everyone can read them.

A photograph illustrating an article. What matters is what it shows in the context of this article, which is rarely a full description of the scene.

A thumbnail next to a headline, both inside the same link. Empty alt, so the link is announced once.

This is why the honest measure of the defect is the count of images with no attribute at all. That number is a fact. Whether the existing alt text is any good is a review, and it takes someone who knows what the page is for.