Image SEO in 2026: Ranking in Google Images, Lens, and AI Overviews

Most image optimization advice stops at file size and load speed. That matters, and our Core Web Vitals guide covers it in depth. But fast images do not get found. In 2026 a single image competes across five distinct Google surfaces, each with its own ranking signals, and a photo that wins on one can win on all of them if you set it up correctly. This guide covers the surfaces, the signals Google actually uses, and the specific changes that move your images from invisible to indexed.

The Five Surfaces Where Your Images Compete

Image search is no longer one place. According to current 2026 image SEO research, a single optimized image can appear across all of these at once.

Surface	Where it appears	Primary ranking signal
Google Images tab	images.google.com	Alt text plus page authority
Image Pack	Carousel in main search results	Relevance plus image quality score
AI Overview thumbnails	Above organic results	ImageObject schema plus attribution
Google Lens	Visual search via camera	Image similarity plus entity matching
Google Discover	Mobile feed	1200px plus width plus engagement

The critical insight is that you optimize once and win on all five. An image with descriptive alt text, ImageObject schema, and a 1200 pixel wide WebP file ranks well on every surface at the same time. An image missing any one of those signals gets capped on at least two surfaces. Industry estimates put Google Images at roughly a fifth of all web searches, and Google Lens now runs over 20 billion visual queries a month, about 20 percent of them shopping-related.

Alt Text Is Still the Number One Signal

Alt text is the single most direct signal Google has about an image's content. It is the only text that is exclusively about the image, as opposed to surrounding body text, which might discuss many topics at once. Get it right and everything else compounds. Get it wrong and no amount of schema fixes it.

The rules that work in 2026:

• Write a specific, descriptive phrase that matches how a user would search for the image
• Keep it under 125 characters, because screen readers cut off at that length and Google follows the same boundary
• No keyword stuffing, no generic labels like "photo" or "image"
• Add context, not just a label

Google's own team has pointed out that "the beach before the chemical spill" beats "a photo of a beach," because people searching images usually want the information attached to the image, not just any picture. Describe what makes this image different from every other image of its subject.

This is also an accessibility matter. The same alt text that helps Google understand your image is read aloud by screen readers to blind users. Writing good alt text is the rare SEO tactic that helps a real audience and a search engine at the same time. Our accessibility-first image guide goes deeper on the overlap between the two.

The Technical Prerequisites Google Requires

Before any ranking signal can work, Google has to index the image. Several technical requirements gate that, and technical image SEO references lay them out clearly.

• Do not block images with robots.txt. A common mistake is blocking the /images/ folder to save crawl budget. You also block Google Images ranking.
• Serve a supported format. Google indexes BMP, GIF, JPEG, PNG, WebP, SVG, and AVIF. JPEG XL is not yet on that list as of August 2026, which is another reason to ship AVIF with a fallback. Convert with the image converter if your pipeline outputs an unsupported format.
• Use a real src, not a CSS background. Google extracts images from <img> src attributes. CSS background-image is not indexed as an image result.
• Match mobile and desktop. Mobile-first indexing means your mobile page has to carry the same images and alt text as the desktop version. Lazy-loaded galleries that never render on mobile lose their images from the index.

Image sitemaps help with discovery but are not a ranking lever. rel=canonical does not apply to images, because Google matches visually similar versions automatically. Do not waste time on either for ranking purposes.

Structured Data Unlocks the Rich Surfaces

Plain alt text gets you into Google Images. Structured data gets you into AI Overviews, Lens shopping results, and the Licensable badge. The schemas that matter most for images:

ImageObject schema tells Google about the image itself: its URL, caption, creator, license, and acquire license page. Adding the license property gates the Licensable badge in Google Images, which matters for stock photographers and anyone selling image rights.

Product schema feeds the Google Lens shopping graph. If you sell physical goods, Product structured data with an image property is what connects your product photos to a shopper pointing their camera at a similar item in a store. Pair clean product photos with the background remover so the product is the only thing in frame.

Recipe, HowTo, and Video schema each carry image properties that surface thumbnails in their respective rich result types. If your content fits one of these, the schema multiplies the surfaces your image can appear on.

Filename and Surrounding Context

Google reads more than the alt attribute. The full set of textual clues it uses to understand an image includes the image URL itself, the alt and title attributes, the paragraph directly above and below the image, the nearest heading, and the page title. A descriptive filename like red-leather-sofa-product-shot.jpg gives Google another signal that IMG_4821.jpg does not.

The surrounding text matters more than people assume. If your alt text says "red leather sofa" but the nearest heading says "Our Office Chairs," Google has to reconcile conflicting signals and usually trusts the page context. Place images near text that describes the same subject, and the two signals reinforce each other.

Format, Dimensions, and Quality

Google ranks images partly on the image itself, not just the page around it. A few physical properties matter.

• Width of at least 1200 pixels for Discover eligibility. Smaller images get filtered out of the mobile feed.
• Modern formats with fallback. AVIF and WebP compress better, which helps your page load and your Core Web Vitals, which in turn helps image ranking. Use the AVIF converter or WebP converter and keep a JPEG fallback.
• High visual quality. Google's image quality score rewards sharp, well-lit, original images over blurry or watermarked ones. If your source is low quality, the image upscaler can help before you publish.
• Unique images. Stock photos that appear on hundreds of sites get suppressed. Original photography ranks better, even if it is less polished.

Visual Search and Google Lens

Visual search is not image SEO with a camera icon. It is a discovery channel where the query is a picture, often a low-quality phone photo with mixed lighting, motion blur, and partial occlusion. The ranking signals differ from text search, and visual search frameworks break down how matching actually works.

Google Lens uses a mix of embedding similarity, entity extraction, OCR on text in the image, surrounding page context, schema markup, and authority signals. You cannot optimize for embedding similarity directly, but you can give Google clean inputs: a clear view of the product, accurate Product schema, descriptive surrounding copy, and a crawlable page. For retailers, this is now a meaningful traffic channel, not a novelty.

A Checklist That Works Across All Five Surfaces

1. Write descriptive alt text under 125 characters for every content image, with context that distinguishes it from generic shots.
2. Use a descriptive filename, not a camera default.
3. Add ImageObject schema, and a license property if you sell image rights.
4. Add Product schema with an image property if you sell physical goods.
5. Serve images at 1200px or wider for Discover eligibility, with srcset for smaller displays.
6. Use AVIF or WebP with a JPEG fallback, served from a real <img> src.
7. Place images near descriptive text and headings that cover the same subject.
8. Compress images so page load helps your Core Web Vitals, which feeds back into image ranking. The image compressor handles this in your browser.
9. Strip sensitive metadata before publishing if your photos contain GPS or personal EXIF data. The EXIF cleaner removes it without touching pixel quality.
10. Do not block images in robots.txt, and make sure mobile renders the same images as desktop.

Common Mistakes

• Alt text that describes the file instead of the subject. "Photo of product 4821" tells Google nothing. "Hand-thrown ceramic mug in matte black, 12 ounce" tells Google exactly what to rank it for.
• Blocking the images folder in robots.txt. You save crawl budget and lose every image ranking. Almost never worth it.
• Lazy-loading above-the-fold images. Lazy-loading the LCP image delays the largest element and hurts both Core Web Vitals and image indexing. Remove loading="lazy" from hero images.
• Serving the same stock photo as competitors. Google suppresses duplicate images. Original photography, even amateur, ranks better than a stock photo on 200 sites.
• Forgetting mobile parity. If your mobile page replaces images with placeholders to save data, Google's mobile-first indexer sees no image and drops it from the index.

FAQ

How long should alt text be for SEO?

Keep it under 125 characters. Screen readers cut off at that length and Google follows the same boundary. Write a specific, descriptive phrase that matches how someone would search for the image, and add context rather than just a label. Avoid keyword stuffing, which Google penalizes and which reads badly to screen-reader users.

Does Google index AVIF and WebP images?

Yes. Google indexes BMP, GIF, JPEG, PNG, WebP, SVG, and AVIF. JPEG XL is not on the supported list as of August 2026, so serve it with a fallback if you use it. Modern formats like AVIF and WebP also improve page load speed, which feeds back into image ranking through Core Web Vitals.

Does image file size affect image search ranking?

Indirectly, yes. Faster-loading pages score better on Core Web Vitals, and Google has confirmed page experience is a ranking signal. A 2 MB hero image slows your page and hurts the ranking of every image on it. Compress with the image compressor before publishing.

How do I get my product images into Google Lens results?

Use Product structured data with an image property, keep the product as the only subject in frame, and host the image on a crawlable page with descriptive surrounding text. Clean product photos on a plain background, which you can produce with the background remover, give Lens the clearest visual signal to match against a shopper's camera photo.

Do image sitemaps improve rankings?

No, but they help with discovery. An image sitemap helps Google find images it might otherwise miss, especially on JavaScript-rendered pages. It is a crawl aid, not a ranking lever. If your images are in plain <img> tags in server-rendered HTML, you likely do not need one.

Related Tools

• Image Compressor - Shrink images for faster pages and better Core Web Vitals
• Image Resizer - Generate the right dimensions for every surface
• Image Converter - Convert to Google-indexed formats
• JPG to AVIF Converter - Serve modern formats with fallbacks
• EXIF Cleaner - Remove GPS and personal metadata before publishing

References

• Google Images SEO: How to Rank in Google Image Search (2026 Guide) - The five surfaces and their ranking signals
• Ranking in Google Images - Technical prerequisites and indexing rules
• Visual Search SEO framework - How Google Lens matching works
• Google Search Central: Image SEO documentation - Official guidance
• Image Search Optimization: Ranking in Google Images and Lens in 2026 - Surface-by-surface signal breakdown

Conclusion

Image SEO in 2026 is less about a single trick and more about giving Google the same clear signals across five surfaces at once. Descriptive alt text, descriptive filenames, ImageObject or Product schema, a 1200 pixel wide modern format, and surrounding text that matches the image get you into Google Images, the Image Pack, AI Overview thumbnails, Lens, and Discover with one pass of work. The technical floor matters too: do not block images in robots.txt, do not lazy-load your hero, and keep mobile parity. Compress everything with the image compressor so your page experience helps your image ranking instead of hurting it, and strip sensitive EXIF data with the EXIF cleaner before you publish.