Image search now has its own number in Search Console. See what the multimodal filter measures and how to prepare your site’s images for Google Lens.
By , founder and lead strategist at Flowup
Direct answer
Image SEO means preparing a site's images to be found and understood by search, including when the question is a photo. Since September 24, 2026, Search Console has separated these searches in the multimodal filter, which brings together Google Lens, Circle to Search and image uploads. The work starts with a quality image in an img element, a descriptive file name and alt text, context on the page and fast loading.
Sources verified on September 30, 2026.
On September 24, 2026, Google began splitting Search Console's "Web" search type into two parts: searches made with text and multimodal searches, in which the question is an image, a photo or a screenshot. According to Search Engine Roundtable, which reported the change the same day, this is new data that was not counted in the Performance report before.
The volume behind this is large. On October 3, 2024, Google's blog reported that people were making nearly 20 billion visual searches a month with Lens, 20% of them shopping-related. On June 17, 2025, Think with Google quoted Lou Wang, co-founder of Lens: more than 1.5 billion people search what they see every month, and Lens grew 65% in one year.
Until now, image search was almost invisible to the site owner. With the new filter, you can see which pages show up when someone photographs a product, a part or a package. This guide explains what the filter measures, what Google recommends for images and what to check in WordPress.
Where to start, depending on your role
The multimodal filter shows the web results triggered by a query made with an image, photo or screenshot. It brings together Google Lens, Circle to Search on Android, image uploads to Search and the "Search this image" option in Chrome, and it is kept separate from searches made with text.
The definition Google uses, quoted by Search Engine Roundtable, is direct: the multimodal type tracks results "triggered by a query that uses an image, photo, or screenshot." The announcement on the Search Central blog introduced the feature in September 2026.
| Search type | What it counts | Shows queries? | What it is for |
|---|---|---|---|
| Web: text-based | Web results for typed or spoken searches | Yes | The usual SEO tracking |
| Web: multimodal | Web results when the question is an image (Lens, Circle to Search, image upload, "Search this image") | No | Seeing which pages show up when someone photographs something related to your business |
| Image | Results in the Google Images tab, for typed searches | Yes | Seeing how images perform as a result, not as a question |
The lack of queries is the most important limit. Search Console Help explains that, because multimodal searches mostly use images, there is no text query to show, and the queries dimension is unavailable for this search type. According to PPC Land, which detailed the filter in a report published on September 24, 2026, Google names queries as unavailable and mentions no other restriction on clicks, impressions, position, pages, countries or devices.
In practice, the page becomes the clue. If a product page gets multimodal impressions, someone photographed something similar to what it shows. If a corporate page shows up, it may be the building facade, the logo or the packaging. It is also worth following the other new Search Console reports, such as the ones for social media, and bringing everything together in your AI visibility dashboard.
When someone photographs an object, Lens looks across the web for pages and images that match what appears in the photo. We did not find a dedicated guide to Lens in the Google Search Central documentation; the public reference is the Google image SEO best practices guide, updated on March 2, 2026.
The guide makes clear what Google needs in order to use an image: find it, understand it and connect it to a useful page. It finds images in the src attribute of the img element, but it does not index images placed as backgrounds with CSS. It understands the image through the alt text, the file name, the surrounding text and structured data. And it chooses quality pages, with the image near the content that explains it.
Our reading
With Lens, the person photographs the real object: the part on the workbench, the package on the shelf, the equipment on the job site. A stock photo competes with thousands of identical copies on the web and does not show your product. The original photo, with the actual product, is the one most likely to match what the person photographed.
The checklist below follows Google's image SEO guide and its image sitemaps documentation. Each item says what to do and why, with the rule from the documentation itself.
| Item | What to do | Why |
|---|---|---|
| Element | Use img with src, not a CSS background image | Google finds images in the src of the img element and does not index CSS images |
| Alt text | Describe the image accurately, in the context of the page | The guide asks for useful, information-rich text, without keyword stuffing |
| File name | A short, descriptive name, such as stainless-steel-ball-valve-2-inch.jpg | Names like IMG00023.JPG say nothing about the image |
| Placement | Image near the text that explains it, on a page about the same subject | Context helps Google understand what the image shows |
| Sizes | srcset for different screens; in picture, always a fallback img | The guide recommends responsive images with a fallback src |
| Format | JPEG, PNG, WebP or AVIF, depending on the case | Google supports BMP, GIF, JPEG, PNG, WebP, SVG and AVIF |
| Structured data | Mark up the product or article with the image property | It connects the image to the entity and can enable features in the results |
| Preferred image | Indicate the main one with og:image or primaryImageOfPage | Google uses these signals to choose the page's image in the results |
| Image sitemap | Use it when images are loaded by JavaScript or sit on a CDN | It helps Google discover what it might not find on its own; up to 1,000 images per page |
| Speed | Compress, serve at the right size and test in PageSpeed Insights | A heavy image slows the page down for people and for Google |
On the sitemap, Google's image sitemaps documentation, updated on December 10, 2025, records that the caption, title, location and license tags were removed. image:image and image:loc are enough. And structured data only helps when it says the same thing the page shows.
Example
A plumbing supplies distributor publishes the photo of a valve with the file name IMG_4471.jpg and the alt text "product". Switch to 316-stainless-steel-ball-valve-2-inch.jpg and "Ball valve in 316 stainless steel, 2 inches, BSP thread, side view", with the spec sheet right below the photo, and the image starts to say what it is, both to search and to people who use a screen reader.
Every type of company has objects that the customer photographs: the part that broke, the package that ran out, the logo seen on a job site. Publishing good photos of those objects, on the right pages, is the most direct way to show up when the question is an image.
| Type of company | What the customer photographs | What to publish |
|---|---|---|
| Manufacturing | The part, the nameplate, the equipment in use | Photos of the product from several angles, with the product code and spec sheet on the same page |
| Distribution and wholesale | The package, the label, the product code | A photo of the current package and of the label, with the code as text on the page |
| E-commerce | The product seen in a store, on social media or in someone's home | Original photos of the product, with color variations and real-life use |
| Services and B2B | The logo, the building facade, a printed piece | Logo and facade on corporate pages, with the company name as text |
In B2B catalogs, this work starts with the structure of the site: each product with its own page, an original photo and technical data as text, as in the B2B e-commerce site of Mantiqueira Distribuidora, built by Flowup.
WordPress 7.1, released on August 19, 2026, changed image processing: compression, resizing and thumbnails are now handled in the browser, by a WebAssembly build of libvips, according to the release notes. The version also added native support for AVIF, HEIC and HDR gain maps.
Speed remains the most sensitive point. According to the HTTP Archive's Web Almanac 2024, an image is the largest visible element (LCP) on 73.3% of pages on mobile and 83.3% on desktop. Even so, 16% of pages on mobile lazy-load that image, which delays the page, and only 15% use fetchpriority="high" on it, a number that grew with WordPress support in 2023.
When the problem is the performance of the whole site, and not just of the images, the path is Flowup's WordPress optimization and Core Web Vitals service, which starts by measuring field data before touching anything.
Always using the same logo, the same colors and consistent product photos helps people recognize the brand when they come across it in an image search. We did not find Google documentation that treats visual consistency as a ranking factor, which is why it appears here as a good practice.
The gain is practical: someone who photographs a logo on a job site or at a trade show wants to reach the right company. A current logo, published on the site with the company name as text, and the same design across profiles and materials make that connection simpler. That is what branding and visual identity work organizes, and what a brand book records for the whole team.
The plan starts with measurement, because the new filter shows where image search already brings visits. Then come the inventory, the fixes and the comparison.
What is not yet known
In short
Image SEO is the work of preparing a site's images to be found and understood by search, both when a person types a question and when they search with a photo. It involves a quality image in an img element, a descriptive file name and alt text, context on the page, structured data, a sitemap when needed and fast loading.
The multimodal filter is a search type in the Search Console Performance report that, since September 24, 2026, has separated the web results triggered by an image, photo or screenshot: Google Lens, Circle to Search, image uploads to Search and "Search this image" in Chrome. It shows pages, countries, devices and dates, but it does not show queries.
No. The Image search type, which already existed in Search Console, counts the results in the Google Images tab, where the person types a query. The multimodal filter counts web results when the question itself is an image, such as a photo taken with Google Lens. They are different audiences and behaviors, and it is worth looking at both.
We did not find a specific guide to Lens in Google's documentation. The reference is Google's image SEO best practices guide: a quality image in an img element, near the text that explains it, with a descriptive file name and alt text, structured data and a fast page. Our reading is that original photos of the real product help more than stock images.
Alt text helps when it describes the image accurately and in the context of the page. Google's guide asks for useful, information-rich alt text, with keywords used naturally, and advises against stuffing the attribute with terms. Alt text also serves people who use a screen reader, which makes it an accessibility requirement, not only a search one.
Google supports both WebP and AVIF, as well as JPEG, PNG, GIF, BMP and SVG, so the format does not change indexing. The choice affects file size and page speed. WordPress 7.1 added native support for AVIF. What matters is serving the image at the right size and not delaying the page's main image.
Not always. According to Google, an image sitemap helps it find images that the crawler might not discover on its own, such as those loaded by JavaScript. Images in a regular img element, on crawlable pages, are usually found without a sitemap. If you use one, each page can list up to 1,000 images.
Sources: Search Engine Roundtable, multimodal filter in Search Console (September 24, 2026); PPC Land, Lens and Circle to Search data without queries (September 24, 2026); Google Search Central, announcement of the multimodal report (September 2026); Search Console Help, Performance report dimensions; Google's blog, Google Lens and ads (October 3, 2024); Think with Google, interview with Lou Wang on visual search (June 17, 2025); Google Search Central, Google image SEO best practices (updated on March 2, 2026) and image sitemaps (updated on December 10, 2025); WordPress.org, WordPress 7.1 (August 19, 2026); HTTP Archive, Web Almanac 2024, Performance chapter (November 11, 2024). Accessed on September 30, 2026.
Image search is no longer invisible. The multimodal filter shows which pages appear when someone points the camera at a part, a package or a logo, and what is missing for the others to appear. Ranking is not enough. Be the answer.
Technical SEO
Find out how Google reads the images and pages of your site
Flowup's free diagnosis shows how search engines and AI assistants read your site. SEO consulting turns what shows up in Search Console into a plan of fixes, with no promise of rankings.
Founder and lead strategist, Flowup Agency
Guto Bertoncini is the founder and lead strategist of Flowup Agency, which he has run since 2011. He is the author of the B.I.N.A. Method, Novo SEO and the Base Informacional Semântica (Semantic Information Base), and leads the agency's SEO for AI, GEO and AEO practice, preparing companies to be found on Google and cited by artificial intelligence platforms. He writes about search and AI on the Flowup blog and on his official website.
