What Search Console's Web: multimodal rows can and cannot tell you

Common belief: Image search is about a quarter of all searches, so that is where the traffic is.

The correction: That quarter-of-all-search figure comes from a February 2018 US clickstream panel

The 22.6% figure in the image-SEO posts comes from SparkToro's February 2018 Jumpshot US clickstream panel, not from Google. One month of US panel data from 2018 is not a reason to reorganize your website in 2026.

Google now names camera-based searching as its own search type, Web: multimodal: Google Lens, Circle to Search on Android, uploaded images in Google Search, and Chrome's right-click "Search this image." Google says the data is rolling out globally and appears when a site receives traffic from those queries, so if the row is in your account, that traffic reached your pages.

Common belief: The Image search type in Search Console already shows me what people photograph.

The correction: Web: multimodal is a split inside Web, not the Image tab

It does not. In Search Console, Image means search traffic from results shown in the Google Images tab, and the Help page lists it as a separate search type from Web: multimodal.

The two new rows sit inside Web. Web: text-based is search traffic originating from text queries entered into the standard Google search bar. Web: multimodal is search traffic from people searching with images instead of typed words.

Includes search traffic when users search using images (such as with a smartphone camera). This includes searches originating from Google Lens, Circle to Search on Android, uploaded images, and Chrome image search.

Search Console Help, Performance report overview

All three definitions sit on the same Help page, behind the Performance report you already open. Someone pointing a phone at your printed flyer and landing on your donation page can register in Web: multimodal while the Image row stays flat, so checking the Images tab for camera traffic reads the wrong row.

Common belief: Once this shows up in my account, I will see what people pointed their camera at.

The correction: Google states that query data is not available for multimodal searches

No. Google says query data is not available for multimodal searches, because those searches mostly use images rather than text. The Queries dimension lists the search terms that led to your site, and a photograph has no search term to list. No filter will produce one.

Read the row as a count of arrivals. Google says it appears when a site receives traffic from those camera-based searches, so it tells you the traffic landed and nothing about what anyone photographed. Stop holding the monthly review open for a search-terms screen that is not coming.

Common belief: I will be able to tell which image on the page earned the click.

The correction: The data lands on the page, not on the photograph

Google's Search Console metrics documentation says that in search analytics data an image is treated as a link to the host page URL, and that Search Console does not distinguish one image from another within a single page. One page, one row, however many photographs sit on it.

The Pages dimension groups data by the final URL linked by a Search result after redirects, with most data assigned to the page's canonical URL. Two URLs that resolve to one canonical appear under the canonical.

So the page is the unit of work. Sort the multimodal rows by page and you have the short list of pages camera-based searches reached.

Common belief: Google has computer vision now, so alt text and captions are legacy work.

The correction: Alt text is the most important attribute Google names for image metadata

Google's image SEO documentation, last updated 2 March 2026, calls alt text the most important attribute for providing image metadata, and says Google reads it together with computer vision and page contents to understand an image. Computer vision appears in that sentence alongside alt text.

• Write alt text on the images that matter: Google names it the most important attribute for image metadata.
• Fill in captions and image titles, which Google says it reads when extracting what an image is about.
• Move each image next to the copy that explains it, since Google recommends placing images near relevant text.
• Rename files where it is easy and skip it otherwise, because Google says filenames give only very light clues.

Common belief: If it is in Search Console it is in the API, so the monthly report will pick it up on its own.

The correction: The split is a report you read, not a field you can export yet

Not yet. The Search Analytics query reference, last updated 11 August 2026, documents type values for discover, googleNews, news, image, video and web, and no separate multimodal type. A pull keyed to type returns web as one figure, so for now you read this split in the interface.

Google added the same Web: text-based and Web: multimodal split to the Generative AI performance report, which counts impressions in AI Overviews and AI Mode rather than clicks. That report groups by Pages, Countries, Dates and Devices, and leaves out Search Labs experiment data. Google's Help page notes that generative AI insights had rolled out to all websites worldwide as of 31 August 2026. A separate article covers what answer engines publish about getting your pages cited.

This week: open both reports, filter to Web: multimodal, and read the rows page by page, since Pages is the grouping both of them give you. Write the page list down before you edit anything.

Read the rows first, then edit the pages they name

The multimodal split answers one question: which pages camera-based searches reached. It will not tell you what anyone photographed, and it will not separate the images sitting on a page. Take the page list out of the report, then spend the hour on alt text, captions and the copy next to the images on those pages. That is the metadata Google's March 2026 documentation says it reads.

The qualification review is free. All we need is your organization's name and website. Start the free quiz