Google Search Console Adds Multimodal Search Reporting: What Visual Search Means for Your SEO
On 24 September Google added multimodal search reporting to the Performance report in Search Console, and it is one of those small interface changes that carries more weight than it looks. The Web search type is now split in two: Text-based and Multimodal. In plain terms, you can finally see how often your pages show up when someone searches with a picture instead of typing words into a box.
If your business depends on organic search, this update deserves half an hour of your attention this week. Let's sink in and look at what changed, what the data can and cannot tell you, and what to actually do with it.
What multimodal search reporting shows
Until now, everything that appeared in Google's main web results lived in a single Web bucket in the Performance report, no matter how the search started. A person typing “waterproof hiking boots” and a person pointing Google Lens at a pair of boots in a shop window were counted together, and you had no way to separate them.
The new filter changes that. When you open the Search type filter, the Web option now offers two choices. Text-based covers searches that used only typed queries. Multimodal covers web search results that were triggered by a query which included an image, a photo or a screenshot. Google's own definition says multimodal tracks results “where an image was used as part of the search”.
The rollout started on 24 September 2026 and is happening globally. There is nothing to switch on. If your site receives this kind of traffic, the new option simply appears in your Performance report, and you can export the data with the usual Export button. Google has not yet said anything about support in the Search Console API or bulk data export, so for now the interface is where you work with it.
Where multimodal searches actually come from
Four behaviours feed the new report. The biggest one is Google Lens, the camera search built into the Google app and most Android phones. The second is Circle to Search on Android, where a user circles anything on their screen to search for it without leaving the app they are in. The third is uploading an image directly into Google Search. And the fourth is the right-click “Search this image” option in Chrome.
Think about what these moments look like in real life. Somebody photographs a chair in a friend's flat to find where to buy it. A teenager circles a jacket in a TikTok screenshot. A homeowner snaps a picture of a broken boiler part before ordering a replacement. None of these people typed a word, yet all of them may have landed on a product page, a guide or a supplier's website.
That is exactly the kind of traffic that has been invisible in Search Console until this week. It skews mobile, it skews younger, and a large share of it carries buying intent, because pictures are how people search when they already have a specific object in front of them.
The numbers are new, so expect your graphs to move
Because multimodal impressions were not previously counted, adding them changes your totals. Some sites will see impressions rise while average CTR drops a little, simply because the report now includes searches it used to ignore. This is a reporting change, not a ranking change, and it is worth warning whoever reads your SEO reports before they spot the bump and assume something happened to the site.
Measurement specialists have asked for this split for years. Glenn Gabe, among others, has long objected to mixed result types sharing one Web category, because it made clean analysis of organic performance harder than it needed to be. The subdivision finally gives analysts a way to track the two behaviours separately across devices, countries and pages.
The catch: you get no query data
There is one significant limitation, and it is better to know it upfront. When you filter by Multimodal, the Queries dimension goes away. Google's documentation explains that these searches mostly use images rather than text, so specific text query data is not available for this traffic.
In practice that means you can see which pages appear in visual searches, how often, in which countries and on which devices, but you cannot see what the user was actually looking at. If a product page gets two thousand multimodal impressions a month, you know visual search matters for it, but you do not know whether people were photographing your product, a competitor's product or something only loosely related.
It is a familiar frustration. Google's generative AI performance reports, which reached all sites at the end of August, still lack click and query data too. And advertisers, notably, do receive inferred intent terms for image-based searches on the paid side. Organic site owners get less. That asymmetry is unlikely to change soon, so the practical move is to build your analysis around the Pages report rather than waiting for queries to arrive.
Why visual search deserves a line in your SEO plan
It would be easy to file this away as a minor reporting tweak. That would be a mistake. Google has said in the past that Lens alone handles around 20 billion searches a month, and that a large share of them relate to shopping. Circle to Search ships on hundreds of millions of Android devices. Visual search stopped being a novelty a while ago; what was missing was any way for a normal business to measure its own slice of it.
Now that the slice is measurable, budget conversations change. If the new filter shows that 8 percent of your web impressions are multimodal, image work moves from “nice to have” to a line item with a number attached. If it shows close to zero for a business that sells physical products, that is a finding too, and probably a fixable one.
Practical takeaways for businesses building digital products
Start by opening the Performance report, applying the Multimodal filter and benchmarking where you stand. Note the share of impressions, the top pages and the device split, and put the same numbers in your monthly report from now on. Without queries, trend lines per page are the most honest signal you will get.
Then look at your images the way Google Lens does. Original photography beats stock, because a stock photo used on three hundred sites gives Google no reason to send the searcher to yours. Product pages should show the actual product from several angles, in decent resolution, with descriptive file names and alt text that says what is in the picture. Product structured data helps Google connect an image to a price, availability and a brand, which is precisely the information a Lens shopper is after.
For teams building web apps and online stores there is an engineering angle as well. Images need to be crawlable and indexable, which means real image URLs rather than CSS backgrounds for anything that matters, no blocking of Googlebot-Image in robots.txt, and product imagery that renders without waiting on user interaction. Lazy loading is fine when it is implemented properly, but an image that only exists after three clicks inside a JavaScript widget does not exist for visual search at all.
And if you work with an agency or an in-house team on SEO, ask them to separate text-based and multimodal in the next report. The split is two clicks away now, and it tends to start useful conversations about where product photos, lifestyle images and diagrams actually earn their keep.
Final notes
Search Console updates rarely make headlines outside the SEO community, but this one gives businesses something they did not have last week: a measurable view of visual search. The data is incomplete, the missing queries sting, and the API question is still open. Even so, seeing the traffic at all beats guessing.
Our advice is simple. Benchmark your multimodal share this week, tidy up your product imagery over the next quarter, and watch the trend. Search is steadily becoming something people do with a camera as much as a keyboard, and Google has just handed you the first proper ruler for it. Use it before your competitors read their own reports.
Comments