Google Search Console Can Now Track Multimodal Search: What It Means for SEO and AI Visibility

A shopper in a café notices a ceramic vase on the next table. They like it, but they have no idea what it's called, who makes it or what to type. A few years ago, the next step would have been a guess in a search bar ("white speckled round vase") and a lot of scrolling. Now they can point their phone at it with Google Lens, circle it on screen with Circle to Search, or upload a photo they took earlier.

That search never produced a keyword in the usual sense. It can still end on a retailer's product page, a maker's website or a local shop's listing. Until this week, though, site owners had no dedicated view in Search Console that separated those image-led visits from everything else.

On September 24, 2026, Google changed that when it announced multimodal Search reporting in Search Console. This article covers what was actually announced, what the new data can and can't tell you, and why the more interesting story is what it says about how customers find businesses. Throughout, we separate what Google has documented from our own analysis and practical recommendations.

Here is what Google has confirmed. The announcement, published on the Google Search Central Blog on September 24, 2026, adds reporting for "web multimodal search" in two places: the Performance report for Search results and the Generative AI performance report. It was co-authored by the product manager leads for Google Lens and for Search Console. In our reading, that joint authorship points to a coordinated visual search and measurement effort rather than a minor interface tweak.

The mechanism is a search type filter. According to Search Console's Performance report documentation, the web search type is now split into two views: "Web: text-based", covering traditional queries typed into the Google search bar, and "Web: multimodal", covering web search results where an image was used as part of the search.

Google says the rollout is global, starting on the day of the announcement, and that the metrics appear in a property's Performance report once the site is receiving traffic from these searches. An empty multimodal view doesn't mean anything is broken. It may simply mean the site isn't yet receiving traffic from these searches. The data can also be exported like the rest of the Performance report.

In Google's reporting, a multimodal search is a web search where an image is used as part of the search. The wording "as part of" is worth noticing. Our reading is that the image may be the entire query, or one element alongside other input. Google names four experiences that feed the new data, listed below with everyday examples.

Two clarifications help avoid misreading the report. First, this is not the same as the existing "Image" search type, which covers Google Images results. The multimodal filter covers web results reached from an image-led search. The two answer different questions: "How do my images perform in Google Images?" versus "Which of my pages do people reach when they search with an image?"

Second, Google's documentation describes a single multimodal search type. It doesn't document a breakdown by entry point, so as things stand you shouldn't expect to separate Lens traffic from Circle to Search or Chrome image searches inside the report.

On the surface, this is a reporting update. Our view is that the filter matters less as a feature and more as a signal. Platforms tend to build dedicated measurement for behaviour they expect site owners to care about, and we read this launch as a sign that Google treats image-led search as a meaningful, ongoing route to websites rather than a novelty. Google didn't publish usage figures in this announcement, and we won't estimate any.

None of this means typed search is fading. Text queries remain central to how people use Google, and nothing in the announcement suggests otherwise. What's changing is the range of starting points. A search can now begin with a camera frame, a screenshot, an object on a shelf, a photo in a chat, or an image on someone else's website.

Each of those starting points carries context that a person might never manage to put into words: colour, shape, material, style, brand marks, setting. That is what makes this kind of contextual search different. A text query only works if the customer knows the vocabulary. An image-led search doesn't need them to. For a business, it means people who don't know your product's name, your category's terminology or even your brand can still reach you, provided your pages can be matched to what they're looking at.

There is also a measurement consequence. Visual discovery has long been easy to talk about and hard to verify from a site owner's side. With a dedicated segment, teams can start testing their assumptions about it with their own data instead of relying on intuition.

Most SEO workflows are built on a simple model: query, then results page, then click. Keyword research estimates demand, pages are aligned to queries, and Search Console reports which queries brought impressions and clicks. It works because the typed query is a readable stand-in for intent.

Multimodal discovery follows a different path: object, image or context, then the search system's interpretation of it, then results, then discovery. The intent is embedded in the image rather than stated. A photo of a sofa might mean "what is this?", "where can I buy it?", "is there a cheaper version?" or "does it come in another colour?". Same image, very different intentions.

The implication, in our analysis, is that keyword-only reporting becomes less complete. Keyword research tools model typed demand; they can't fully represent demand that begins with a photograph. A business could see flat keyword numbers while picking up discovery through image-led searches that no keyword report was ever going to capture.

That doesn't make keyword research obsolete. It makes it necessary but partial. The practical shift is from matching phrases towards making each page unambiguous about what it is: which product, which place, which service, for whom. That clarity serves both typed and visual searches.

Start with what's documented. In the Performance report for Search results, choosing "Web: multimodal" in the search type filter segments the report to web searches where an image was part of the search, separately from typed ones. The Performance report's standard metrics are clicks, impressions, click-through rate and average position. Google's documentation doesn't list any exceptions for the multimodal view, but confirm what your own property shows.

In the Generative AI performance report, the same filter applies. That report shows impressions in generative AI features on Google Search, currently AI Overviews and AI Mode, grouped by page, country, device and date. Per Google's help page for the report, the metric it shows is impressions.

Now for what isn't documented. Google's announcement and help pages don't describe a way to see the image a person searched with, or which of your images was matched. As noted above, they don't document a split by Lens, Circle to Search, uploads or Chrome. And they don't explain what, if anything, appears in the Queries dimension for multimodal searches. Check that in your own property rather than assuming. Some multimodal searches may contain no typed words at all, so there may be nothing resembling a traditional query to report.

In our view, that changes how the data should be analysed. For multimodal traffic, the page, not the query, becomes the most dependable unit of analysis. Instead of asking "which keywords brought these visits?", the more useful question is "which pages are image-led searches reaching, and what do those pages have in common?" Our suggested approach:

Any website with pages that correspond to something a person could photograph or screenshot has a stake in this. The businesses most exposed are those whose products or places are visual and hard to describe precisely in words.

That doesn't leave other industries unaffected. An industrial supplier whose parts get photographed on a job site to find a replacement, or a B2B brand whose equipment appears at trade events, is just as open to image-led search. The deciding question is whether your offering can be seen, not which category you're in. The list below shows where the connection is most obvious:

To be clear about the facts first: Google's rollout is global, and nothing in the announcement is specific to Oman, the UAE or the wider GCC. We're also not aware of reliable public data on how widely visual search is used in the region, so we won't guess.

What we can say is that several of the region's most important sectors are strongly visual: hospitality and tourism, restaurants and cafés, retail and malls, real estate, and automotive. Think of a visitor photographing a restaurant's façade in Muscat, a guest screenshotting a hotel lobby in Dubai from a friend's story, or a buyer saving an image of a villa from a social media advert. Each of those is a plausible starting point for an image-led search.

There is also a language angle. An image has no language, but the page it leads to does. For bilingual businesses, that makes clear, well-structured content in both Arabic and English, with business details that match across both versions, more important, not less.

For local businesses, the same applies to Google Business Profile photos and listing details. Genuine, current images of the premises, products and team, alongside consistent name, address and category information, are already part of sound local SEO practice, and they give image-led searches something accurate to connect to.

Many websites still treat images as decoration: stock photography, generic banners, text baked into graphics, filenames like IMG_4821.jpg. In an image-led search, the image and the page around it may be exactly what connects a customer's photo to your business. In our view, that turns visual assets into discovery assets.

Google didn't publish a Lens-specific optimisation checklist alongside this announcement. The most relevant official reference remains Google's image SEO best practices. Google describes them as common to its visual discovery features, such as Google Images and Discover. They aren't Lens-specific, but they are the closest official guidance. The points below combine that guidance with our own practical recommendations:

The factual link is straightforward. Google put the multimodal filter inside the Generative AI performance report too, so site owners can see impressions in AI Overviews and AI Mode that came from image-led searches. Google is connecting the two in its own reporting.

Our broader reading is that visual search and generative AI search are parts of the same shift: search systems that interpret more context, handle more formats and increasingly compose answers instead of only listing links. The multimodal filter's presence in the Generative AI report indicates that an image-led search can lead to a generative AI feature, not only a list of results.

Google's generative AI optimization guide is explicit that these features are rooted in its core Search ranking and quality systems, and that SEO best practices remain relevant. It says that if you already follow its image SEO and video SEO guidance, you're already optimizing for generative AI search. It also states that structured data isn't required for generative AI search and that there's no special schema.org markup to add, while recommending structured data as part of an overall SEO strategy.

So there is no single "AI SEO ranking factor", and schema markup doesn't place a business in AI answers by itself. What tends to help, in our experience, is information that is machine-understandable and trustworthy: crawlable pages, clear descriptions of what a product, service or place is, consistent entity details, images that are clearly tied to that information, and credible independent evidence. We've explored the wider picture in our look at how AI is reshaping customer discovery.

We've also written about businesses ranking well but missing from AI answers. Multimodal search adds another layer to that gap: a business can have strong text visibility and still be poorly represented when the search begins with an image.

The framework below is practical, not a promise. None of these steps guarantees visibility in Google Lens, AI Overviews, AI Mode or any other system. They improve the chances that your content can be found, understood and matched when someone searches with an image.

Several of the checks overlap with standard technical SEO work, which is good news: the effort isn't duplicated. The same groundwork serves typed search, visual search and AI features alike.

One caution on measurement: impressions and clicks from image-led searches only matter if they lead somewhere. As with any channel, more traffic isn't the same as more business, so connect what Search Console shows to enquiries, bookings or sales in your analytics.

SEO isn't becoming irrelevant. Google's own guidance says the opposite, and nothing in this announcement suggests typed queries are going away. What is broadening is discovery. A business can now be found through a typed keyword, a conversational AI prompt, a photo, a screenshot or a circled object in a video.

We think a more useful goal than "ranking for keywords" is what we'd call discovery visibility. It's a working term, not an industry standard, and it means asking whether your information can be found, interpreted and trusted across search interfaces and formats, not only whether a page ranks for a phrase.

In practice, that pulls together work that often sits in separate teams: SEO, content, photography, product data, local listings and analytics. In our view, visual search favours businesses where all of those tell the same, accurate story.

The new filter won't answer every question. It does answer one that Search Console couldn't answer directly before: whether people who searched with an image ended up on your site. A sensible first step is a modest one. Open the Performance report, switch the search type to "Web: multimodal", note what's there, and start treating your images as part of how customers find you, not just how your pages look.