Search Console Now Reports Multimodal Search. What Does It Tell You?

Google Just Added Multimodal Search to Search Console. What Does It Actually Tell You?

As of September 2026, Search Console reports how your pages perform when someone searches with a camera instead of a keyboard. Google announced it on 24 September and it covers Lens, Circle to Search, image uploads, and the Chrome right click option. It is a new filter, not a new report.

Our expectation is that most B2B software sites will find very little in this filter, because the surfaces it covers are driven by people pointing a camera at physical things. That does not make the change unimportant. It makes it a measurement change worth understanding before somebody builds a strategy on top of it.

Here is what Google shipped, what it means, and what we would and would not do about it.

What Exactly Did Google Announce?

Google introduced "Search Console reporting for web multimodal search, in the report for performance on Search results as well as in the report for Generative AI features." The post is dated Thursday, 24 September 2026, and is credited to Harsh Kharbanda, a Product Manager Lead on Google Lens, and Moshe Samet, a Product Manager Lead on Search Console.

The data covers a specific set of entry points. Google lists "searches with Lens, Circle to Search on Android, image uploads to Google Search, and the Chrome right-click 'Search this image' feature." Those are the four surfaces. Anything else visual is not in scope.

Access is simple. Google says to "use the new multimodal search type filter in the performance reports" and that you can "Click 'Export' to download the data and analyze it in your other tools." The rollout is "global starting today", and Google adds that "you will start seeing these metrics in your Performance report if your site is receiving traffic from these queries."

Why Does a Search Type Filter Matter More Than It Sounds?

A search type filter changes what your existing numbers mean. Before this, multimodal traffic was either missing or blended into other buckets, and neither is good. Once Google splits it out, your web totals and your conversion rates can shift without anything changing on your site.

This is the familiar failure mode around any Search Console segmentation change. Somebody notices a line move in a monthly report, assumes a ranking event, and spends a week investigating what turns out to be a reporting boundary. Knowing the filter landed on 24 September saves that week.

The second reason it matters is that filters are how Google tells you what it considers a distinct surface. Google does not build a search type for something it treats as an afterthought. We read this the way we read the arrival of the AI features reporting, which we covered in our piece on the Search Console AI performance report.

How Big Is Visual Search, Really?

Large in consumer contexts and small in most B2B ones. Google says on its own blog that "every month, people use Lens for nearly 20 billion visual searches, and 20 percent of all Lens searches are shopping-related." Google also says its Shopping Graph holds "more than 45 billion products."

Read those two numbers together and the shape becomes clear. Lens is enormous, and a fifth of it is people shopping. That is a retail and consumer behaviour, driven by pointing a phone at an object. It is not how a procurement lead evaluates a data platform.

So the honest framing for a B2B software company is that this is a real surface where you are probably not present, and probably should not spend heavily to be. The value of the new filter for you is diagnostic, not promotional.

Who Should Actually Care About This Report?

Anyone who sells a physical object, publishes photography, or runs a site where an image is the product. Manufacturers, hardware companies, marketplaces, travel and hospitality brands, and any business whose product can be photographed in the wild have a genuine reason to open this filter.

There is a second group that is easy to miss. Companies with physical locations, packaging, signage, or printed materials get photographed by customers without ever intending it. If somebody points a camera at your box, at your building, or at a page of your printed guide, that is now a measurable search.

The group that can safely ignore it is pure software with no physical presence and no original imagery. If your site is screenshots and stock photography, this filter will stay near zero and that is the correct outcome, not a failure to optimise.

What Should You Do in the First Week?

Open the performance report, apply the multimodal search type filter, and set the date range as far back as it will go. Write down the number. That single figure tells you whether this is a channel or a rounding error for your site, and it takes about two minutes.

If the number is meaningful, export it. Google explicitly supports export here, and the Search Console interface is a poor place to do real analysis. Pull the queries and pages into a spreadsheet and look at which pages are being reached, because that is usually more surprising than the query list.

If the number is near zero, note the date and move on. Do not restructure your image strategy on the basis of a report that has nothing in it. We walk through how we read the rest of the platform in our Search Console guide.

Does This Change How You Should Handle Images?

Only at the margins, and only for sites with real photography. The fundamentals of image work did not change on 24 September. Descriptive file names, accurate alt text, correct dimensions, modern formats, and fast loading were already the job, and they remain the job.

What changes slightly is the case for original photography over stock. A camera search matches against the thing in front of the lens. Stock imagery that appears on a thousand other sites gives a matching system no reason to prefer you, while a photograph of your actual product does.

The accessibility argument still does most of the work here. Good alt text serves screen reader users first and machine understanding second, and we made that case in our piece on image SEO for Google Images. If the visual search benefit arrives too, treat it as a bonus rather than the reason.

Is This Connected to AI Search, or Separate?

Connected, and Google said so in the announcement. The reporting appears "in the report for performance on Search results as well as in the report for Generative AI features", which places multimodal data alongside the AI surfaces rather than in a corner of its own.

That placement is the interesting part. It suggests Google sees a camera query and a conversational query as two doors into the same generative experience, not as separate products. A photograph is just another prompt.

For anyone planning for answer engines, the practical read is that your content needs to be retrievable from more than one kind of input. Text you can read, images you can match, and structure a machine can parse are all the same job seen from different angles.

What Would Make Us Change Our Advice?

Volume in the B2B segment. If multimodal impressions on software sites ever reach a level that competes with ordinary web search, the advice changes from "measure it" to "plan for it". Nothing Google published on 24 September suggests that is the case today, and we would rather say so than manufacture urgency.

The second trigger would be Google exposing what the camera actually matched. Today you get queries and pages. If Google ever shows which image on the page pulled the result, image level optimisation becomes a real discipline instead of a guess. That is the report we would genuinely like to have.

Until either happens, this is a good measurement improvement and a poor strategy driver. Both of those can be true at once.

How Should You Explain This to Your Team?

Say that Google added a way to see camera driven searches, that it started on 24 September 2026, and that your number is either interesting or near zero. Put the actual figure in the message. A report with a number in it ends the speculation that a report without one starts.

Then set an expectation about the totals. If your Search Console web numbers look slightly different this month, part of that may be segmentation rather than performance. Flagging that in advance is much easier than explaining it after somebody has drawn a chart.

Finally, put a reminder in the calendar for a quarter out. Surfaces that start near zero sometimes do not stay there, and a filter you checked once and forgot is the same as a filter you never opened.

What Comes Next for Visual and AI Reporting?

We expect more segmentation, not less. Google has now split out AI features and multimodal search inside the same performance report, and each split makes the underlying totals more honest and the month over month comparisons more delicate. That trade is worth it, but it needs handling.

The teams that will cope best are the ones that already record what changed and when. A dated note next to a chart is worth more than a dashboard, because the dashboard cannot tell you that Google moved a boundary on a Thursday in September.

If you want a second pair of eyes on what your Search Console data is really saying, or on whether your images are doing any work at all, we are happy to walk through it. Find us at phoenix.studio and bring the export.