Search Console Now Splits Text and Multimodal Searches

Have you ever wondered if anyone finds your site by pointing a phone camera at something? Search Console can now tell you. On 24 September 2026, Google split web search performance data into text-based and multimodal searches. The new search console multimodal view shows how your pages perform when people search with an image through Google Lens, Circle to Search and similar tools.

In this guide I explain what counts as a multimodal search and where to find the new filter. I also cover how to read data that has no queries in it, which sites are likely to see the most activity, and how to make your images and pages easier for Lens to match.

What Google Changed on 24 September

Google announced the change in a Search Central blog post, Announcing web multimodal Search performance reporting in Search Console. The rollout is global and started the same day.

Here are the confirmed points:

• A new search type. The Performance report’s search type filter now offers web data as two options: text-based and multimodal.
• Two reports. Multimodal data shows in the Performance report for Search results and in the Generative AI features report.
• Which searches count. Google lists Google Lens, Circle to Search on Android, images uploaded to Google Search, and Chrome’s right-click “Search this image” option.
• No queries. The queries dimension is not available for multimodal data. These searches mostly start from an image, not typed words.
• New data, not moved data. Google has said this traffic was not counted in Search Console before. Your existing web numbers have not been moved into a new group. This is extra visibility you could not see before.
• Only where traffic exists. Metrics show up for sites that really get traffic from these searches. Plenty of sites will see very little, or nothing.

What a Multimodal Search Actually Is

A multimodal search uses more than one kind of input. In this report, the key input is an image. Someone takes a photo of a lamp in a café, circles a jacket in an Instagram video, uploads a screenshot of a plant, or right-clicks a picture in Chrome. Sometimes they add words, such as “where to buy” or “cheaper version”. But the image is still the starting point.

Google’s definition is simple. Text-based covers searches typed into the normal search bar. Multimodal covers web results where an image was used as part of the search.

Multimodal Is Not the Same as Google Images

This confuses a lot of people. The Image search type in Search Console reports results shown in the Google Images tab, where people type words and browse pictures. Multimodal reports the web results Google shows after someone searches with a picture.

So a product photo can earn traffic in two different ways. It can show in Google Images for a typed search. And the page it sits on can show as a web result when someone points Lens at the same product in a shop window.

Where to Find the Search Console Multimodal Filter

The filter sits inside the existing Performance report. You don’t need to verify anything new or turn anything on.

1. Open Search Console and choose your property.
2. Go to Performance and select Search results.
3. Click the Search type filter at the top of the report.
4. Under web, choose Multimodal instead of Text-based, then apply.
5. Set the date range you want. Look at the chart, then the tabs below it.

To see both side by side, open two browser tabs with one search type in each, using the same date range. That is how I do it. The same multimodal option is also in the Generative AI features report, which I covered in my guide to Search Console’s generative AI reports.

One practical note. This traffic was never counted before. So think of the date the data first appears for your property as a starting line, not a jump in performance. Add a note to your own reports so nobody later reads it as a sudden win.

How to Read the Multimodal Report

The report looks familiar, but having no queries changes how you use it. Treat it as a page report first.

Start With the Pages Tab

With no queries to look at, the Pages tab becomes your main view. It tells you which URLs Google showed as web results after someone searched with an image.

When I open this report, I look at the top 10 to 20 pages and ask:

• What is the main image on each page? A product shot, a finished dish, a diagram, a photo of a place?
• Is the image original, or a stock or manufacturer photo that many other sites also use?
• Is the page the most useful place to land for someone who just took a photo of that thing?

Patterns appear quickly. An online furniture shop might find that its product pages with lifestyle photos show up, but its category pages don’t. A recipe site might find its finished-dish photos bring in people who took a photo of a meal.

Clicks, Impressions, CTR and Position

The usual metrics apply to whatever Search Console records for this search type. Read them with the shape of Lens results in mind. People are often comparing several similar-looking results on a phone. So position and click-through rate won’t always behave the way they do for typed searches.

Don’t compare multimodal CTR with your text CTR. Compare multimodal with itself over time, and compare pages within the multimodal view.

Use Devices and Countries

Lens and Circle to Search are mostly used on phones, so expect mobile to lead. If a page gets multimodal impressions mostly on mobile, check that page on a real phone. Slow image loading, pop-ups and awkward product galleries hurt more here. The visitor came from a camera, often while standing in a shop or on a street.

Countries can show you markets where people search visually for your products, even if most of your typed-search traffic comes from somewhere else.

Compare Against Text-Based Results

Export the Pages tab for both search types over the same date range and put them in one sheet. Then look for three groups:

• Pages strong in both. These are your best all-rounders. Protect them when you redesign.
• Pages strong in multimodal only. Their images are doing work their text isn’t. These pages could use better titles, descriptions and on-page copy.
• Pages strong in text only. If they sell or describe something visual, the images may be weak, generic or hard for Google to link to the page. In my experience, this group usually has the easiest wins.

Google’s announcement mentions the Export button for this data. It says nothing about the Search Console API or bulk data export. So check your connected tools before you assume multimodal numbers flow into them. If you report through a Looker Studio SEO dashboard, confirm which search type your connector pulls and label it clearly.

The Generative AI Features View

Multimodal data also appears in the Generative AI features report. That helps when someone searches with an image and gets an AI-written answer that cites pages. The limits of that report still apply, so expect less detail there than in the main Search results report.

Why Your Multimodal Report Might Be Empty or Tiny

Many sites will open the filter and see very little. That isn’t always a problem.

• Your content isn’t visual. A legal advice site, a SaaS help centre or a news analysis blog rarely matches what people take photos of.
• Your images are generic. If you use the same supplier photos as dozens of other stores, Google has many near-identical pages to choose from. This is the most common weakness I see on e-commerce sites.
• Your images aren’t indexable. Images loaded as CSS backgrounds, hidden behind scripts or blocked by robots.txt are hard for Google to use.
• Low volume. If only a few people search with images for what you offer, the chart will show scattered days with no data. That reflects real demand, not a fault.

An empty report is still useful. It’s your baseline. Check it again after you improve your key images.

Which Sites Benefit Most

Visual search is about recognising things. Sites that sell, describe or locate physical things have the most to gain.

E-Commerce Stores

This is the clear winner. People take photos of shoes on a stranger, a sofa in a hotel lobby, a spare part with a worn label, or a product in a competitor’s store. Lens then shows shopping and web results that match.

Stores with their own photos, several angles and clear product details are in a better spot than stores using only the manufacturer’s single white-background shot. Product structured data also helps Google understand the price, stock and identity of the item on the page.

Local Businesses

People use Lens on shopfronts, menus, signs and landmarks. A restaurant, a gallery, a hardware shop or a tourist attraction can show up when someone points a camera at it.

Local businesses often get much of this visibility through their Business Profile, not their website. And Search Console only reports website performance. Still, a site with real photos of the premises, its products and its signs gives Google more to match. This matters most for businesses with distinctive interiors, dishes or products.

Visual Products and Visual Topics

Beyond shops, multimodal searches suit any topic where the question starts with “what is this?” or “how do I do this?”:

• Plant, garden and pest identification sites
• Recipe sites, where people take a photo of a dish
• Fashion, interiors and beauty publishers
• DIY, repair and spare parts sites, where people take a photo of a broken part
• Art, antiques, collectibles and vehicles
• Education sites covering diagrams, maths problems or homework, which people often screenshot

Circle to Search also brings in screenshots and on-screen content. A page explaining a chart, a meme or a product seen in a video can earn traffic this way.

How to Optimise Images and Pages for Google Lens

Google hasn’t published separate ranking rules for Lens. What it has published is its image SEO best practices for Google Images. Those basics help Google find and understand your images and link them to the right page. I covered the basics in my image SEO guide. Here’s how I would apply them with Lens in mind.

1. Use Original, Clear Photos of the Real Thing

Lens matches what the camera sees. Clear, well-lit photos of the actual product, dish or place give it the best chance. Show several angles, close-ups of special details, and the item in use. A shopper who takes a photo of a chair from the side should be able to match your side-view image. If I could change only one thing on a product site, it would be this.

2. Make Images Indexable

Google can’t use what it can’t find:

• Use standard HTML <img> elements with a src attribute. Google doesn’t index CSS background images.
• Use a supported format such as JPEG, PNG, WebP or AVIF. Keep the file extension matching the file type.
• Don’t block image folders or your CDN in robots.txt.
• Add images to an image sitemap if they load in ways Google might miss.
• If you use srcset or <picture>, keep a fallback src.

3. Describe Images With alt Text and File Names

Write alt text (a short text description of the image) that says what the image shows, in plain language. “Walnut mid-century armchair with green wool seat, side view” is far more useful than “chair” or a list of keywords. Use short, clear file names, such as walnut-armchair-green-side.jpg instead of IMG_4821.jpg.

4. Put Images Next to the Text That Explains Them

Google uses the page around an image to understand it. Place the main image close to the product name, the description and the key facts. A Lens visitor wants to check quickly that they found the right thing. So the name, price or answer should be visible without much scrolling.

5. Add Structured Data

On product pages, use Product markup with the image, name, price and stock. Use Recipe markup for recipes and LocalBusiness markup for physical locations. My schema markup guide covers which types to use and how to test them.

6. Make the Landing Page Work on a Phone

Most multimodal visitors are on mobile. Fast image loading, a gallery that works with a thumb, and no annoying pop-ups all help. Use sharp images, but compress them so the page doesn’t slow down.

7. Write Titles and Descriptions That Confirm the Match

In a Lens result, the title and snippet help the searcher decide if your page shows the same thing they took a photo of. Specific titles with the product name, model, colour or dish name do this better than vague, clever ones.

What This Report Can’t Tell You

Be honest with clients and managers about the limits:

• No queries. You won’t know if someone took a photo of your exact product or something similar.
• No history before the data starts. You can’t compare with last year, because this traffic wasn’t counted before.
• No revenue link. To estimate value, join the page-level export with sales or lead data for the same URLs. That shows a connection, not proof.
• Not all visual discovery. Traffic from Google Images is reported separately, and Business Profile activity isn’t in Search Console at all.

If you put this in a client report, I suggest keeping it to a short trend line and a list of the top pages. My guide to SEO reporting for clients explains how to show new metrics without overselling them.

Conclusion

The search console multimodal filter finally shows how your pages perform when people search with a camera or a screenshot instead of a keyboard. It’s new data. So start by recording a baseline, read the Pages tab instead of looking for queries, and compare it with your text-based results.

If you sell or describe anything visual, your next step is to check the images on your top multimodal and text pages. Replace generic photos with original ones, fix alt text and file names, and make sure the images are indexable. Then check the report again in a few weeks to see what moved.