Image SEO: Name, Describe and Compress Your Photos, Then Find Out Who Copied Them

Google cannot see your photo the way a customer does

Google's image SEO documentation says it combines computer vision with the text around an image to work out what it shows: the alt attribute, the file name, the page title and the copy nearby. The pixels tell it there is a deck. The words tell it the deck is cedar, in Helotes, built after a storm took the old one. Only the words help someone searching for exactly that job.

Before any of that, Google has to find the file. Its documentation says it reads images from the src attribute of an HTML img element, including one inside a picture element, and does not index images set as CSS backgrounds. A hero photo dropped in through a stylesheet looks identical to a visitor and does not exist for Google Images. If an image matters for search, put it in the HTML as an img with a real src.

File names and alt text that describe instead of chant

Google's guidance on file names is one line: short and descriptive beats a camera's default name or a generic one like image1.jpg. It also asks that file names be translated when a page is localized, which matters for a Texas business with Spanish pages. Alt text gets more attention. Google calls it the most important attribute for describing an image, and its example climbs from missing alt, to a single word, to a short phrase saying what the subject is doing.

Two rules sit behind the table. Write what the picture shows, as you would describe it to a customer on the phone, and stop. Google warns against filling alt text with keywords, and a string of city names is the most common version we see. Second, alt text exists for people who cannot see the image. MDN's reference notes that an empty alt marks an image as decoration that screen readers may skip, which is the right call for flourishes and spacers. For inline SVG, Google says to use a title element instead. How images fit with headings, copy and links on the rest of the page is covered in on-page SEO.

WebP, AVIF and the JPEG you still keep around

Google Search indexes BMP, GIF, JPEG, PNG, WebP, SVG and AVIF. AVIF was the last to join: Google's Search Central blog announced support in August 2024, with nothing special required to get the files indexed. Format choice is mostly a speed decision, not a ranking trick.

Google adds one housekeeping rule: the file extension should match the actual file type. A PNG renamed to .jpg confuses more than it saves. Most modern platforms convert formats for you. Next.js, which we build on, resizes images through its image component and serves WebP by default, with AVIF a one-line setting, and our notes on moving WordPress images to Next.js cover what changes in a migration. If WordPress refuses the upload in the first place, see the fix for the HTTP error when uploading images.

Sizes and lazy loading without hiding photos from Googlebot

Google's image documentation says images are often the largest contributor to page size, which is why they decide so many Core Web Vitals results. The fixes are mechanical, and every one of them is checkable.

1. 1

   Resize to the display size

   A photo shown 800 pixels wide should not ship at 4,000. Export a few widths and let srcset pick, so phones get the small one.

2. 2

   Set width and height on every img

   MDN and Google's web.dev both recommend it: the browser reserves the space before the file arrives, so the layout does not jump. That is Cumulative Layout Shift avoided for free.

3. 3

   Never lazy-load the first big image

   web.dev says not to lazy-load images likely to be in view when the page opens, especially the Largest Contentful Paint image. Load it normally, and consider fetchpriority="high" on it.

4. 4

   Lazy-load the rest the right way

   Use the browser's loading="lazy" attribute or IntersectionObserver. Google's lazy-loading guidance says content must load when it enters the viewport, because Googlebot does not scroll or click.

5. 5

   Reuse the same URL

   When one photo appears on several pages, Google asks that you reference the same image URL so it can be cached, rather than uploading copies.

6. 6

   Check what Google rendered

   Run the page through URL Inspection in Search Console and confirm the image URLs appear in the src attributes of the rendered HTML.

Image weight is usually the cheapest speed win on a small business site. Our guide to improving Core Web Vitals covers the other metrics, and our Core Web Vitals service does the work when the theme or the plugins fight back. Images injected by scripts carry the same rendering risks as any client-side content; see JavaScript SEO.

Image sitemaps, CDNs and the picture Google should pick

Google's image sitemap documentation describes them as a way to point out images it might otherwise miss, such as those loaded by JavaScript. You can add image tags to an existing sitemap or keep a separate file; Google says either is fine. Each page entry can list up to 1,000 images, and only two tags still matter: image:image and image:loc. Caption, title, geo location and license tags were removed from Google's documentation. Our guide to XML sitemaps covers the file itself.

• Images on a CDN are fine. The sitemap can list image URLs on another domain, and Google encourages verifying that domain in Search Console. Your robots.txt must not block the images.
• Tell Google which image represents the page. Its documentation names three ways: primaryImageOfPage in schema.org markup, an image attached to the page's main entity, or the og:image tag. Our meta tags guide covers og:image.
• Pick a real photo for that job. Google prefers relevant, high-resolution images and advises against logos, images full of text and extreme aspect ratios.
• Product images belong in structured data. Google's image documentation says the image property is required for a page to be eligible for a badge and rich result in Google Images. See schema markup and our guide to product pages.

Google Images, Lens and searches that start with a photo

People increasingly search with a camera. Search Console's Performance report now separates multimodal searches, started from an image through Google Lens or Circle to Search, from text searches, a change Google announced in September 2026. Those rows carry no query text, because the search was a picture. Our Google Search Console guide shows where the filter lives. Google's guidance on AI features in Search, updated in 2026, also tells site owners to add high-quality images and video, since those features can bring them into answers.

Two features reward getting the details right. Image metadata, through structured data or IPTC fields embedded in the file, can show the creator, credit line and copyright in Google Images, and a license URL makes an image eligible for the Licensable badge. Google does not guarantee either appears. And About this image, opened from Lens or Circle to Search, shows how other sites use and describe a picture and any metadata about how it was made. For a business, the lesson is to keep your own photos original and labeled. Photos of real work are also the strongest evidence of experience, which our E-E-A-T guide explains.

How to reverse image search your own photos

A reverse image search starts from a picture instead of words and returns pages that show the same or a similar image. It is the quickest way to learn whether a competitor lifted your project photos, whether a directory is using an old logo, or whether a blog used your chart without credit. Here is how, per Google's own help pages.

Search your five best project photos, your logo and any original chart or diagram. Cropped or edited copies may still surface as similar images, so scan beyond the first row. Then repeat every quarter or so; copies appear quietly.

Found a copy? Ask for a link before you ask for a takedown

Most people who use your photo without permission are not villains. A supplier featured your job, a local blog wrote about your neighborhood, a trade association needed a picture. Each of those is a site that already likes your work, which makes it a warm prospect for a credit link. A polite note asking them to credit the photo with a link to the project page often works, and it is one of the methods in our guide to how to get backlinks.

A competitor passing your work off as theirs is a different matter. Google says it complies with notices under the Digital Millennium Copyright Act and takes them through its legal removal help center, which asks for the exact URLs involved. Write to the site owner and its host first, keep screenshots and your original files with their dates, and talk to a lawyer before anything formal. DataCram is not a law firm, so we can help you find the copies and document them, not advise on the claim.

If you would rather have all of this done, image work is part of every DataCram SEO program: renaming and describing the library, converting and resizing files, sitemaps, and a periodic check for copies that could become links. Programs start at $625 a month, and most local businesses land between $625 and $2,250, per our SEO pricing. To see where your images are slowing pages down today, run the free site audit, and find the rest of the subject in the SEO guide.