Images: file names, alt text and metadata

The file name is a small clue#

Google says that the file name gives a very light clue about what the image shows. A short, descriptive name is better than a name from the camera: black-kitten.jpg says more than IMG00023.JPG. Names like image1.jpg or pic.gif should be avoided [1].

A file name is therefore worth getting right when the image is uploaded. It is not worth renaming a thousand images that already exist. A new name gives the image a new address, and the old address then has to lead on.

The alt text and the text around the image matter more#

• The alt text is the most important fact about an image. Describe what the image shows in ordinary words. Do not stuff it with keywords [1].
• Placement. Put the image near text that is about the same thing [1].
• Real image tags. Google does not index images that are only a background in CSS. Images that should be found need an <img> [1].
• The page's title and text are also used when Google describes the image in the results [1].

EXIF, IPTC and GPS: what Google actually reads#

Google Images reads a few fields from the IPTC standard: Creator (the photographer), Credit Line (who should be credited), Copyright Notice, Web Statement of Rights and Licensor URL [2] [3]. Web Statement of Rights is a link to the image's licence. It is required for the Licensable badge in Google Images [2]. If structured data and IPTC say different things, the structured data applies [2].

Google suggests that you can remove metadata to make the file smaller, but that you should keep the copyright details. In some countries it can be illegal to remove them [2].

Myths about images#

• Myth: EXIF and GPS in the image give a better position. No. Google's Martin Splitt said in September 2024 that Google does not use EXIF for ranking [4]. John Mueller has written that geotagging images is not needed for SEO, and tests have shown no clear effect [5].
• GPS in a published image is a privacy matter. Anyone who downloads the image can see where it was taken, for example a home. It does not affect Google at all.
• Myth: keywords in the file name lift the page. Google calls the file name a very light clue [1]. Five keywords in a file name do no good.
• Myth: copyright in the image gives a higher position. Google describes the IPTC fields as details that are shown in Google Images. Nowhere does it say that they affect the position [2].

Size, format and loading#

• Format. Google can read BMP, GIF, JPEG, PNG, WebP, SVG and AVIF. The file extension must match the file type [1].
• The right size for the screen. With srcset the browser can choose a smaller image on mobile. Always give an ordinary src as well [1].
• Width and height. With width and height on the image, the browser keeps space for it, so that the page does not jump when the image arrives [6].
• Lazy loading only further down. Images further down the page can load when the visitor scrolls there. Images on the first screen should load at once [7] [8].
• The main image first. With fetchpriority="high" the largest image is fetched before other things, and it shows earlier [9]. More about speed is in Speed.

Images in the sitemap#

A sitemap can list the images on every page with <image:image> and <image:loc>. That helps Google find images it might otherwise miss. Google has removed the tags for caption, title, location and licence from its documentation. A page can have up to 1,000 images in the sitemap [10].

This is what we do#

• Descriptive file names. Images on sites we build are named after the caption, industry, town and company name, for example rormokare-jarvso-badrum.jpg (plumber, Järvsö, bathroom). At most five words, no repetitions, å, ä and ö become a, a and o. Files that already exist keep their names, so old addresses work.
• Clean metadata. We remove the GPS position, the camera's serial number, the owner's name and other EXIF from every image we upload. We keep the colour profile and copyright details that are already in the image.
• Credit, not made-up copyright. The company's name is written as the Credit Line. Photographer, copyright and licence link we only write when you have given them. We never claim that anyone owns an image.
• The right way up. A phone photo where the camera has stored that it should be rotated is shown the right way up, also in the smaller copies for mobile.
• Fast images. Width and height, smaller copies for mobile, lazy loading further down and the highest priority for the image on the first screen.
• Image sitemap. The sitemap lists the company's own photos under the page where they are shown.
• Copies never rename images. When we move an existing website, every image keeps its address (see Switch without losses).
• In the analysis you see images with meaningless names, images without alt text or size, very heavy images (over 500 kB), images with a GPS position and, for information, photos without details of the photographer or copyright.