Search Console now counts camera searches. How to act on a report with no queries

KEY TAKEAWAYS

1. The report gives pages, not queries. Rerun the search yourself with each top page's own image.
2. Google says the multimodal numbers are new data. Check your first date before comparing periods.
3. Matching is visual first. Indexable, original images and complete product data come before alt text.
4. Do not judge a page group on fewer than about 30 multimodal clicks.

On 24 September 2026 Google added a search type called Web: multimodal to the Search Console Performance report. It counts searches that start with an image. It shows pages, countries, devices and dates, and it shows no queries. Most guides published since then describe the filter and tell you to fix alt text. This article covers what the numbers mean, where the sources disagree, and a method for acting on a report that cannot say what anyone searched for.

What does the multimodal filter count?

Google's help page now splits the Web search type in two. Web: text-based covers text queries typed into the standard search bar. Web: multimodal covers searches made with an image, and the page names four entry points: Google Lens, Circle to Search on Android, uploaded images and Chrome image search.

This is not the Image search type. That one still means results in the Google Images tab.

Search Engine Watch reports that the filter shows clicks, impressions, CTR and average position. The same split exists in the generative AI performance report, which Google's help page says reached all sites on 31 August 2026 and which shows impressions only. Treat those impressions with the caution we set out for AI visibility sampling.

Three limits matter:

• No queries. Google's dimensions page says text query data is not available for this traffic because the searches mostly use images, and the queries dimension is switched off for this search type.
• No split by entry point. Search Engine Watch notes that Lens, Circle to Search, uploads and Chrome are reported as one number.
• No API value yet. The Search Analytics API reference, last updated on 11 August 2026, lists web, image, video, news, discover and googleNews as types. There is no multimodal value, so for now the data leaves Search Console through the Export button.

Is this new traffic or old traffic with a new label?

The sources disagree, and the answer changes how you read your trend lines.

Search Engine Roundtable recorded the exchange. Dave Smart asked on Bluesky whether these impressions used to sit inside the Web total. He had seen impressions and clicks fall on a few properties from the 10th, which matched the new data. Google's John Mueller first said he thought the searches had not been reported before. After checking with the team he wrote: "the data wasn't previously in the counts, the report has new data."

The team's answer is the best evidence available, and we would take it as the default. Two things stop it being the whole story. Search Engine Watch found one property with multimodal data reaching back to 10 September, two weeks before the filter appeared, and Google has announced no common start date. And Google's data anomalies page had no entry for the change when we read it on 6 October.

The timing adds noise. The filter arrived in the same week Google began its September 2026 spam update, according to Search Engine Roundtable's weekly recap. A step in your Web numbers that week has more than one possible cause. Our guide to diagnosing a drop during the spam update covers the other causes.

So do not assume. Find the first date with multimodal data in your own property. Then check whether text-based clicks moved on that date.

Google's numbers are not your numbers

Google's figures for visual search are large. In October 2024 it said Lens handled nearly 20 billion visual searches a month and that 20% were shopping-related. In June 2025 a Think with Google interview put Lens at more than 1.5 billion monthly users and 65% growth year over year. Both are internal figures with no published method.

Neither tells you what share of your traffic starts with an image. Digital Applied points out that Google published no traffic share with the launch. The filter is the first place a site owner can measure it. If your multimodal row is close to empty, that is a finding too, and the right response is to do nothing.

Does alt text decide who appears?

Nearly every guide to the new filter ends with alt text and filenames. The evidence for that is thin.

Google's image SEO documentation does call alt text important. It says Google uses it together with computer vision and the contents of the page to understand an image. That sentence describes how Google understands an indexed image. It does not say a camera search is matched on text.

Gianluca Fiorelli's Lens guide for Advanced Web Ranking, published in August 2026, argues the opposite for photo searches: Lens reads the object, not your alt text. He separates exact matches, which find the same image elsewhere on the web, from visual matches, which find things that look alike. That is his inference from how the product behaves, not a statement from Google. SerpApi's documentation does confirm that Lens returns exact matches, visual matches and products as separate result types.

The one large study is old. Backlinko analysed 65,388 Lens results in June 2020. Only 11.4% of result images had alt text matching the search label, while 32.5% of pages had a matching keyword in the title tag. It is correlational and six years out of date. Mueller added in January 2025, as Search Engine Journal reported, that choosing alt text is not primarily an SEO decision.

Our reading: matching is visual first, and page text then shapes how the result is understood and labelled. Write alt text for accessibility. Spend the SEO effort on whether the image can be indexed, whether it is yours alone, and whether the product data behind it is complete.

The missing query matters less than it looks

In text search the query is a string you cannot guess. In image search the query is a picture, and often it is your own picture, screenshotted or photographed. You can rerun it.

Google's help page describes the route in Chrome. Right-click an image and choose to search it with Google Lens. Do that for the main image on a page and you see roughly what an image searcher sees. If other sites appear in exact matches, other sites carry the same photo. If you appear only among lookalikes, you are competing on resemblance.

One caution on reading the metrics. Google's help page on impressions and position has rules for image blocks, carousels, AI Overviews and AI Mode. It has no section for Lens results, so average position in the multimodal view is undocumented. Do not compare multimodal CTR with text CTR. Compare pages with each other inside the multimodal view.

What each pattern means: a decision table

The last row follows advice from Lou Wang, co-founder of Lens, in the Think with Google interview. He said keeping older products and images online is useful, with a pointer to a newer version or a similar item in stock.

A seven-step method

1. Find your start date. Set search type to Web: multimodal and note the first day with data.
2. Export both views. Same date range, pages tab, once for multimodal and once for text-based.
3. Group pages by template. Product, category, guide, tool and so on. Add up impressions and clicks per group.
4. Apply a floor. A click count of n carries a 95% range of roughly plus or minus 2 divided by the square root of n. That is our calculation. At 10 clicks the range is 63% either way, at 30 it is 37%. Below about 30 clicks, read impressions only.
5. Rank pages by multimodal impressions and take the top 20.
6. Run Lens on each page's main image. Record whether you appear in exact matches, who else does, and what the visual matches look like.
7. Fix by pattern and log the date. Export again every week, because the API cannot do it for you.

For step 7, Google's own documents set the baseline. CSS background images are not indexed. Googlebot does not scroll or click, so a lazy-loaded image must appear in the rendered HTML. Rules for Googlebot-Image apply wherever images show in Search. Our crawler access guide explains how such rules stack. For products, the merchant listing documentation requires an image property, recommends several images of at least 50,000 pixels in area, and requires image URLs to be crawlable and indexable. Merchant Center recommends images of around 1500 by 1500 pixels or larger, with the product filling 75% to 90% of the frame.

A worked example with invented numbers

This retailer is hypothetical. It sells furniture and exports four weeks of data.

Multimodal is 8.2% of all impressions and 3.0% of all clicks. Category pages have 10 clicks, so their CTR is unreadable. Product pages have 300, a range of about 12% either way, and guides have 180, about 15%. The gap between 0.5% and 2.0% is real.

The top 20 product pages hold 45,000 of the 60,000 product impressions, or 75%. Running Lens on their main images shows the same supplier photo on a dozen other shops. The retailer is one seller among many in exact matches, which fits high impressions and few clicks. The guides are different. Their original diagrams return the retailer first.

The plan is narrow: original photography and complete identifiers for 20 products, and more original diagrams in guides. It skips a site-wide alt text rewrite.

Where this stops

A click in this report is not a sale. Join the exported pages to your analytics before calling the traffic valuable.

The report may also change. Google Ads documentation already says its search terms report may show a best approximation of intent for journeys such as Lens searches, as Search Engine Roundtable reported in May 2026, so a query-like field is not impossible. Google has promised nothing. If you want to follow how we track brand presence across search surfaces, join the waitlist.

Questions

What does Web multimodal mean in Search Console?

It is a search type in the Performance report for searches where the user searched with an image. Google's help page lists Google Lens, Circle to Search on Android, uploaded images and Chrome image search. It is separate from the Image search type, which covers the Google Images tab.

Why are there no queries in the multimodal report?

Google's help page says these searches mostly use images instead of text, so text query data is not available, and the queries dimension is switched off for this search type. You can still group the data by page, country, device and date.

Did the multimodal filter change my existing Web totals?

Google says no. John Mueller checked with the Search Console team and said the data was not in the counts before. One practitioner saw impressions and clicks fall on a few properties from 10 September, so check your own first date before comparing periods.

Can I pull multimodal data through the Search Console API?

Not from what the documentation shows. The Search Analytics API reference, last updated on 11 August 2026, lists web, image, video, news, discover and googleNews as type values and no multimodal value. Use the Export button and keep the files.

Sources

1. Google Search Console Help, Performance report (Search results), Overview and basic setup support.google.com
2. Google Search Console Help, Performance report (Search results), Dimensions and data groupings support.google.com
3. Google Search Console Help, What are impressions, position and clicks support.google.com
4. Google Search Console Help, Generative AI performance report (Search) support.google.com
5. Google Search Console Help, Data anomalies in Search Console support.google.com
6. Google for Developers, Search Analytics query API reference developers.google.com
7. Google Search Central, Google image SEO best practices developers.google.com
8. Google Search Central, Merchant listing structured data developers.google.com
9. Google Search Central, Fix lazy-loaded content developers.google.com
10. Google Search Central, Google's common crawlers developers.google.com
11. Google Merchant Center Help, Image link support.google.com
12. Google Search Help, Search with an image on Google support.google.com
13. Google, Google Lens now offers shopping product details blog.google
14. Think with Google, Google Lens co-founder on visual search trends business.google.com
15. Search Engine Roundtable, Google Search Console Performance Report Gains Multimodal Search Type Filter seroundtable.com
16. Search Engine Roundtable, Google Ads Search Terms Report May Show Best Approximation Of User Intent seroundtable.com
17. Search Engine Roundtable, Video recap for 25 September 2026 seroundtable.com
18. Search Engine Journal, Google Search Console Adds Multimodal Filter For Image-Based Searches searchenginejournal.com
19. Search Engine Journal, Google Confirms Alt Text Is Not Primarily An SEO Decision searchenginejournal.com
20. Search Engine Watch, How to use Search Console's new multimodal search filter searchenginewatch.com
21. PPC Land, Google Search Console gains Lens and Circle to Search data, without queries ppc.land
22. Advanced Web Ranking, Google Lens, Visual Search Explained advancedwebranking.com
23. Backlinko, Google Lens Study, Results From 65,388 Visual Searches backlinko.com
24. SerpApi, Google Lens API documentation serpapi.com
25. Digital Applied, Search Console Now Separates Visual Search Traffic digitalapplied.com
26. Scale Xpert, Google Search Console Now Shows Image-Led Web Traffic scale-xpert.com

NEXT STEP

See where your brand stands.

Your Digital Authority Score, who AI recommends for your target questions, and the placements that close each gap. Your first report is free.

Get your free report →