Skip to content
search-engine-optimizedby Netsolution

Images and image search

An image is not understood on its own: the engine combines the alternative text, computer vision algorithms and the page's content to work out its subject. The file name gives only a very light hint, and the caption is among the elements read.

An image is not understood on its own

That is the sentence to keep before any optimisation, and it is documented as such.

The subject of an image is worked out by combining the alternative text, computer vision algorithms and the page's content.

officialSource : Google Search Central, “Google image SEO best practices”

Three supports, of which two belong to the page and not to the image. An illustration dropped into a text that does not mention it therefore gives up two thirds of what would allow it to be interpreted, even if it is sharp, light and perfectly named. That is the opposite of the common reflex, which is to work on the file and leave the page as it stands.

A black-and-white film contact sheet: strips of negatives printed side by side, most of the frames crossed out in blue chinagraph pencil.
Every frame exists, and nearly all are crossed out. A contact sheet is for sorting before printing — that is the work alternative text and a caption do for an image published online.Si Griffiths — CC BY-SA 3.0 · Wikimedia Commons, relevé le

Alternative text, and the rank it is given

Alternative text is presented as the most important attribute for supplying metadata about an image.

officialSource : Google Search Central, “Google image SEO best practices”

“The most important”: the documentation ranks things itself, which is rare, and that is enough to settle an order of priority. A site where half the images have no alternative text has no reason to worry about file names.

That leaves what to write in it. The most reliable criterion is also the simplest: alternative text describes the image for someone who cannot see it. That formulation comes from accessibility and it settles nearly every case — it produces a description, not a list of words, and a description is precisely what the attribute is asked for. That the same sentence serves two audiences is no coincidence: both need the same thing.

Writing descriptive alternative text on images that carry meaning

hypothesis

Effort
moderate
Time to effect
weeks
Risk
none
Durability
structural

A purely decorative image is the exception: it has nothing to describe, and inventing a description for it adds noise for a screen reader and for an engine alike. The question to ask is “what is lost if the image disappears?” If the answer is nothing, there is nothing to write.

The file name, and its proper place

The file name gives only very light clues about the subject of an image: the documentation advises short but descriptive names.

officialSource : Google Search Central, “Google image SEO best practices”

The words the documentation uses are “very light”, and they are worth more than any amount of editorial caution: it remains a signal, it is weak, and it does not justify a retroactive renaming project. On images still to come, on the other hand, naming them properly costs nothing — it is the kind of gesture that only has value when applied at the moment of creation.

What the page brings to the image

Information about the subject of an image is extracted from the page's content, captions and image titles included.

officialSource : Google Search Central, “Google image SEO best practices”

A caption is therefore not an ornament. It is one of the few places where you can spell out what an image shows and why it is there, which alternative text alone never says well — it describes, it does not argue. Placement counts too, for a plain reason of common sense: an image surrounded by the text commenting on it is easier to interpret than an image relegated to the bottom of the page or to a carousel that separates it from its point.

Formats and discovery

Images referenced in the source attribute of an image tag are supported in the BMP, GIF, JPEG, PNG, WebP, SVG and AVIF formats.

officialSource : Google Search Central, “Google image SEO best practices”

One detail of that sentence counts as much as the list: support is described for images referenced in the source attribute of an image tag. An illustration displayed only by a style rule does not take that path. If an image carries meaning it belongs to the page’s content, and therefore to an image tag; if it belongs to the styling, the stylesheet is exactly where it should be.

An image sitemap allows the address of images that would not otherwise have been discovered to be declared, and it permits, unlike an ordinary sitemap, addresses belonging to other domains.

officialSource : Google Search Central, “Google image SEO best practices”

The second half is a genuine exception to how sitemaps work, and it has a concrete use: a site whose visuals are served from a separate domain can declare them anyway. Like any sitemap, this one helps discovery and guarantees nothing beyond that.

What does not work

Fill in every image's alternative text with the page's keyword: it is a prime slot for reinforcing your relevance.

The attribute exists to describe the image, and the subject is worked out by combining that description with computer vision and the page's content. A keyword repeated across every image describes none of them, contradicts the other two signals and degrades accessibility on the way.

Two other habits deserve flagging. The first is weight: a page whose illustrations dominate the loading time pays for its images in reading experience, and that is a trade-off to examine in its own right. The second is more insidious — illustrating out of obligation. An image that brings nothing does not become useful because someone wrote it good alternative text; it becomes a correctly described element nobody needed.

Sources

  1. Google image SEO best practices — Google Search Centralaccessed on
  2. Learn about sitemaps — Google Search Centralaccessed on

Block revalidated on

Frequently asked questions

Is alternative text for search or for accessibility?
For both, and without conflict: it describes the image for whoever cannot see it, human or machine. The documentation presents it as the most important attribute for describing an image, and text written for a screen reader already fills that role correctly.
Should you rename your image files?
It helps but it is minor: the documentation itself calls the clues a file name gives very light. Renaming a whole library is therefore a low-yield project, whereas writing the missing alternative texts acts on the element presented as the most important.
Can an image be understood on its own by the engine?
Not on its own: the subject is worked out by combining alternative text, computer vision and the page's content. An image dropped into a page that does not talk about it therefore loses two of those three supports, whatever the quality of the visual itself.
Which formats are supported?
The documentation lists the formats supported when the image is referenced in the source attribute of an image tag: BMP, GIF, JPEG, PNG, WebP, SVG and AVIF. An image displayed only by a style rule does not go through that attribute, and so falls outside that frame.