Search Console Multimodal Search: How to Track Google Lens Traffic

Search Console multimodal search is a new search type in the Performance report. Set the Search type filter to Web: multimodal and you see clicks, impressions, pages, countries and devices for web results people reached by searching with an image, through Google Lens, Circle to Search, image uploads or Chrome. You do not see queries.

Google announced the feature on its Search Central Blog on September 24, 2026 and is rolling it out globally. This guide is part of the SEO hub and walks through one analysis, from the filter to a page level action, as of October 2026. The main recommendation: treat multimodal data as a page level signal, because the query is missing by design.

Is multimodal traffic new data or traffic moved out of the web totals?

Multimodal traffic is new data. Google's John Mueller said the multimodal impressions were not previously part of the Search Console counts, so Web: multimodal adds traffic that was invisible before September 24, 2026. It was not carved out of the old web totals, and a drop in Web: text-based needs another explanation.

When Dave Smart asked whether these impressions used to sit inside the web totals, Mueller answered after checking with the team: "the data wasn't previously in the counts, the report has new data."

Multimodal search is a search where the input is an image, photo or screenshot, sometimes combined with text. Google Lens is Google's visual search tool that finds information about what a camera or an image shows. Circle to Search is the Android feature for searching anything on the screen. Google lists four entry points: Google Lens, Circle to Search on Android, image uploads to Google Search and the Chrome right-click "Search this image" feature, and a site only sees metrics once it receives this traffic, according to Google's announcement.

The Performance report help page defines Web: multimodal as traffic "when users search using images (such as with a smartphone camera)." The report tooltip, quoted by Search Engine Roundtable, adds: "Text-only queries are tracked as Web text-based."

How does Web: multimodal compare with text-based, Image and Discover data?

Web: multimodal describes how the search started, from a picture, while Web: text-based covers typed queries and the Image search type describes where the result appeared, in the Google Images tab. Discover is not search and has its own report. One product page can collect impressions in several views, with different data in each.

Typing "green velvet armchair" in the Images tab creates an Image impression; pointing Lens at that armchair creates a multimodal one. The Generative AI performance report (Search) covers AI Overviews and AI Mode, counts impressions only and has the same Web: multimodal filter.

View in Search Console	What triggers a row	Queries available	Clicks available	Best use
Performance report, Web: multimodal	Web result for a search that started from an image	No	Yes	Pages that match what people photograph
Performance report, Web: text-based	Web result for a typed text query	Yes	Yes	Classic keyword and snippet work
Performance report, Image	Result shown in the Google Images tab	Yes	Yes	Image ranking for typed queries
Generative AI report, Web: multimodal	Link to your site in AI Overviews or AI Mode for an image search	No	No, impressions only	Visibility in AI answers for camera searches
Performance report (Discover)	Content shown in the Discover feed	No queries in Discover	Yes	Feed reach, separate from search

What does a worked multimodal analysis look like, from filter to page list?

A worked multimodal analysis fits in one Search Console session: set the Search type filter to Web: multimodal, group by page, compare against Web: text-based and Image, then export the views and join them on the page URL. The result is a page list sorted by multimodal impressions.

The Search type filter cannot be removed, but its Compare option puts two search types side by side, according to Google's advanced filtering help page. The usual 1,000 row limit applies, and "~" or "-" values become zeros in an export.

Which page signals stand in for the missing queries?

Pages, the images on them, devices, countries and dates stand in for the missing queries. Search Console hides query text for multimodal traffic because the input is mostly a picture, so the question shifts from "what did people type" to "which image on this page matches what people photograph", and that answer comes from the page itself.

Google explains the gap on its dimensions help page: "Because multimodal searches mostly use images rather than text, specific text query data isn't available for this traffic." Mueller summed it up on Bluesky: you do not see the photo, "but you can see which of your pages showed up & how often."

Group the multimodal view by device and country as well. Circle to Search runs on Android, so a strong mobile share hints at camera and screen searches rather than image uploads.

Multimodal Page Triage: which action fits which page pattern?

The Multimodal Page Triage is the author's own rule, not Google guidance: it maps five patterns in the multimodal and text-based data to a likely meaning and one action. It turns the exported page list into a short to-do list, starting with pages that collect multimodal impressions but few clicks.

What the page shows	Likely meaning	Action
High multimodal impressions, low multimodal CTR	Google matches the image, but the result does not convince	Check the title, the main image and whether price or availability is visible
High multimodal impressions, no text-based impressions	People find it by picture, not by name	Add the product or object name, model and attributes to the page text and alt text
Multimodal and Image impressions on the same page	The images rank well in both views	Protect them: keep file URLs stable during redesigns
No multimodal impressions on a visual product page	Google may not connect the images to the page	Check that images use HTML img elements, carry alt text and are in an image sitemap
Multimodal impressions on an unexpected page	A generic image matches many photos	Decide whether that traffic fits the page intent before optimizing it

Which image SEO basics does Google document for multimodal search?

Google has not published separate guidance for multimodal search, so the image SEO basics in Google Search Central are the documented levers: HTML img elements, descriptive alt text, images near relevant text, image sitemaps and supported formats. For ecommerce sites they decide which product pages Lens returns; for publishers, which articles surface for photos.

The closest official guidance is its image SEO best practices:

• Use HTML image elements. Google finds images in the src attribute of an img element and does not index CSS background images.
• Write descriptive alt text. Google uses alt text "along with computer vision algorithms and the contents of the page" to understand an image.
• Place images near relevant text. Captions, titles and the surrounding copy tell Google what the image shows.
• Submit an image sitemap. Image sitemaps may list image URLs on other domains, such as a CDN; our sitemap index guide covers how to submit and verify sitemaps.
• Use a supported format. Google Search supports BMP, GIF, JPEG, PNG, WebP, SVG and AVIF.

Google said in an October 3, 2024 post that Lens handles nearly 20 billion visual searches every month and that 20 percent of all Lens searches are shopping-related (2024 figures). The new filter measures the part that reaches your own pages.

Should you build dashboards on multimodal data yet?

Multimodal data is not ready for trend dashboards yet. As of October 2026 the Search Console API has no multimodal type, Google has not said from which date the series starts, and the Performance report has had reporting delays since the launch. Use exports for page analysis now and hold off on week over week charts.

The Search Analytics API reference lists the types web, image, video, news, discover and googleNews, with no multimodal value. Google's announcement points to the Export button instead.

PPC Land noted that Search Live and image prompts in AI Mode are not named in Google's list and that Google has not said from which date the multimodal series starts. Google has also not said whether searches that pair an image with typed words always count as multimodal.

Search Engine Roundtable reported on October 11, 2026 that the Performance report was more than 64 hours behind, after a spam update, and that Google had not commented at the time. Check the "Last updated" date before you compare weeks.

For AI visibility beyond Google, see our overview of AI visibility tools, and for AI referral clicks in GA4, the guide to Gemini referral traffic.

Common mistakes when reading Search Console multimodal search data

The common mistakes with multimodal data come from treating it like text search data or like the Image search type. Each one leads to the wrong conclusion about a page, and each has a fix that only needs the Performance report and a spreadsheet.

• Looking for keywords that do not exist. The Queries tab is unavailable for this search type. Work from pages and the images on them instead.
• Reading a text-based drop as traffic moved to multimodal. Google says the multimodal data is new and was never in the counts. Look for another cause, such as an update or a delay.
• Mixing up Image and multimodal. The Image type is the Images tab; multimodal is a search that started from a picture. Compare them side by side instead of adding them up.
• Comparing periods across the launch date. Data before September 24, 2026 may not exist for this type. Start comparisons after the launch.
• Optimizing every image for the camera. A page that wins multimodal impressions for an unrelated picture does not need more of that traffic. Check whether it fits the page intent first.