On September 24, 2026, Google added a new “Multimodal” option to the Web search-type filter in Search Console. It separates web searches that began with an image — Google Lens, Circle to Search on Android, image uploads to Google Search, and Chrome’s right-click “Search this image” — from searches that began with typed words, in both the Performance report for Search results and the report for Generative AI features. You can see which of your pages surfaced through image-led searches, how often, and on which devices and in which countries. What you cannot see: the query text (there usually isn’t one) or the actual image someone searched with. Below is what the filter measures, where to find it, and how to use it without misreading your own data.
What exactly changed
Google announced the change in a Search Central blog post on September 24, 2026, written by Harsh Kharbanda, Product Manager Lead for Google Lens, and Moshe Samet, Product Manager Lead for Search Console. The announcement introduced Search Console reporting for web multimodal search in two places: the report for performance on Search results and the report for Generative AI features.
In practice, the change looks like this: open the Performance report, open the Search type filter, and the Web option now splits into two — Text-based and Multimodal. Google’s updated Performance report help documentation defines the two as text-based traffic from queries typed into the standard search bar, and multimodal, which “includes web search results where an image was used as part of the search.” The in-product tooltip phrases it slightly differently: multimodal “tracks search results triggered by a query that uses an image, photo or screenshot.”
Google began rolling the new reporting out globally on the announcement date. One detail to set expectations early: the data appears only for properties that are actually receiving traffic from these searches. An empty multimodal view is not a bug — it means the site isn’t getting image-led traffic yet. The rollout reached accounts gradually rather than all at once, so a property that lacks the option on day one may gain it later.
Where to find the filter
The steps, per Google’s announcement and the updated help documentation:
- Sign in to Search Console and select the property you want to inspect.
- Open Performance > Search results from the left navigation.
- At the top of the report, open the Search type filter.
- Under Web, choose Web: multimodal (or Web: text-based to see only typed queries) and apply.
- Use the Pages, Countries, Devices, and Dates tabs below the chart to break the numbers down.
The same split is available in the report for Generative AI features, which Google says reached all sites on August 31, 2026. One caveat on that report: it counts impressions, not clicks, so don’t expect click-through analysis there. To work with the data outside Search Console, Google points to the Export button at the top of the report — you can download the filtered numbers and analyze them in spreadsheets or BI tools. Note that as of the announcement, the Search Analytics API reference (last updated August 11, 2026) did not list a separate multimodal value for its type parameter, so manual export is the documented path for pulling this data out.
What counts as “multimodal”
Google’s announcement names four entry points that feed the new reporting:
- Google Lens — including searches started with a smartphone camera, where someone points their phone at an object and searches visually.
- Circle to Search on Android — circling, highlighting, or tapping an object on screen to search it.
- Image uploads to Google Search — a photo or screenshot uploaded directly into Search.
- Chrome’s right-click “Search this image” feature — searching any image found while browsing.
What all four have in common: the search begins with an image rather than with words. That’s the defining line. A multimodal search can still produce classic web results — that’s why this lives under the Web search type rather than the Images search type. It is also why the filter is not a duplicate of the existing “Images” search type, which tracks the Google Images vertical. Two searches for the same product can now be reported separately depending on whether the user typed or pointed a camera.
What the filter does not show
The most important limitation is the queries dimension. When you select the Multimodal search type, the Queries tab is unavailable. Google’s dimensions help documentation explains why: “Because multimodal searches mostly use images rather than text, specific text query data isn’t available for this traffic.” When someone points a camera at a pair of shoes or circles a product on screen, there is simply no typed phrase behind the search for Search Console to report.
Equally important: you cannot see the image that triggered the search. As Google’s John Mueller put it when the feature launched: you don’t see the photo, but you can see which of your pages showed up and how often. The report tells you which pages were discovered through image-led search — not what people searched for, and not what the original image contained.
What you do get for the multimodal slice: clicks, impressions, and average position in the Search results report, broken down by page, country, device, and date. That is enough to answer real business questions — which product pages get camera-driven discovery, where that demand concentrates geographically, and whether it skews mobile — without pretending to keyword-level precision that the data can’t support.
The totals question: was this traffic in your Web numbers before?
This is where coverage has been messiest, so here is what is actually established. When the feature launched, some site owners reported corresponding drops in their Web impressions and clicks around the time the new filter appeared, and asked Google whether multimodal traffic had previously been grouped into Web totals. John Mueller replied on the discussion thread that he would double-check, adding that he thought the traffic was not reported at all before. Trade coverage has since reported that Google clarified the multimodal traffic was “not previously in the counts,” while at least one consultant issued a correction noting that the complete reporting history had not been fully established by Google.
The practical takeaway doesn’t depend on which version you believe: do not treat Web totals from before September 24 as directly comparable to Web totals after it. Before running any before-and-after analysis, pull both sides of the comparison fresh with the same search-type setting — text-based alone against text-based alone, or multimodal alone against multimodal alone. Mixing a pre-rollout “Web” export with a post-rollout “Web: text-based” view is the most likely way to invent a traffic drop that never happened.
A related wrinkle: Google started its September 2026 spam update on the very same day, September 24, with a stated rollout window of up to two weeks. If you see movement in late-September data, the multimodal split and the spam update are two separate possible causes sitting on top of each other. Check both slices before assigning a drop to either.
What it means for publishers and SEOs
For most text-first publishers, the multimodal slice will be small — in one documented launch-window case, it amounted to around 0.2% of Web impressions over September 10–21 on a single property. That’s one site’s observation, not a benchmark, and image-heavy sites should expect a larger share. The value of the filter is not the size of the number today but what it lets you track over time: visual discovery has a different context than typed search, and now it has its own line in the reporting instead of being folded into one number.
The sites with the most to gain are the ones people already search by pointing cameras at things:
- Ecommerce and product catalogs — shoppers photograph products, packaging, or barcodes in stores and search them with Lens. A product page earning multimodal clicks is reaching buyers at the point of visual intent.
- Local businesses — someone photographing a storefront, a menu, or a dish. The Countries and Devices breakdowns matter here, since this behavior skews mobile.
- Publishers with strong photography — recipes, DIY guides, fashion, and travel content where the image itself is the discovery hook.
For these sites, the filter answers questions that were previously unanswerable: which pages are being found through image-led searches, whether that share is growing, and how those pages’ text-based and multimodal performance compare. If a product page ranks well for typed queries but earns zero multimodal impressions while competitors’ equivalents get them, that’s a visual-content problem — weaker photography, missing product images near relevant text, or images that crawlers can’t index properly.
That last point is worth spelling out: the filter measures discovery, it doesn’t fix it. Standard image SEO still applies to the visual path — crawlable image files with descriptive filenames and alt text, high-quality photos placed near relevant text. CSS background images are not how Google indexes images, so decorative styling won’t carry a page here.
How to put the filter to work this week
A short routine for getting value out of the new reporting:
- Check availability first. Open the search-type filter in one or two of your largest properties. If Multimodal isn’t there yet, note the date and check again later — the rollout is gradual.
- Establish a baseline, not a verdict. Look at the last 28 days of multimodal clicks and impressions per property. Compare against the same period’s text-based slice. You’re measuring share of discovery, not chasing a target number.
- Sort pages by multimodal impressions. The Pages tab shows which URLs are actually being found through image-led searches. Look for patterns: product pages, specific categories, or individual images that carry the weight.
- Check the country and device splits. Multimodal discovery is inherently mobile-first — confirm that assumption against your own data rather than assuming it.
- Export and keep both slices. Download text-based and multimodal exports separately so future comparisons stay apples-to-apples, and keep the Exports labeled with the search-type setting you used.
- Repeat the split in the Generative AI features report. Image-led searches that surface your site inside AI Overviews or AI Mode are readable separately there too — impressions only, but worth tracking as generative surfaces grow.
Common mistakes to avoid
- Confusing Multimodal with the Images search type. The Images search type tracks the Google Images vertical. Multimodal tracks web results where the search began with an image. They measure different things.
- Waiting for query data. The Queries tab will not appear under the multimodal filter. Treat this as page-level analytics — which pages surfaced, how often, where — and plan content around that.
- Reading an empty filter as broken. No multimodal data usually means no multimodal traffic, not a configuration error. It is itself a signal about how little visual discovery your site currently gets.
- Comparing old “Web” exports to new “Web: text-based” numbers. As covered above, the two aren’t the same denominator. Always compare like with like.
- Expecting API parity on day one. If you automate reporting through the Search Analytics API, the multimodal dimension wasn’t in the API reference at launch — manual export is the documented route until that changes.
Keep reading
This article is part of DroidOrbit’s coverage of Google’s Search Console updates. Continue with the companion pieces in this cluster:
- Search Console Multi-Country View: One Traffic Report for All Countries — how Google is merging per-country reporting into a single view.
- Google Wins AI Overviews Lawsuit Dismissal: What Publishers Do Now — what the dismissal of the publisher suits means for sites relying on search traffic.
- Frase vs Surfer SEO: Which SEO Content Tool Wins in 2026? — comparing two content-optimization tools for the SEO work that follows measurement.
Frequently asked questions
What is the Search Console multimodal filter?
It’s a search-type option added to Search Console’s Performance reports on September 24, 2026. Under the Web search type, you can now split traffic into Text-based (typed queries) and Multimodal (searches where an image was used as part of the search). It covers Google Lens, Circle to Search on Android, image uploads to Google Search, and Chrome’s right-click “Search this image” feature.
Where is the multimodal filter in Search Console?
Open Performance > Search results, then open the Search type filter at the top of the report. Under Web, select “Web: multimodal.” The same split is available in the report for Generative AI features. If you don’t see the option, your property may not have the rollout yet, or the site may not be receiving image-led traffic.
Does the multimodal filter show search queries?
No. The Queries tab is unavailable when the Multimodal search type is selected, because image-led searches usually have no text query behind them. You can still see clicks, impressions, and average position, broken down by page, country, device, and date.
Can I see which image someone searched with?
No. Search Console shows which of your pages appeared in image-led searches and how often — it does not reveal the submitted photo, screenshot, or the contents of the image.
Was multimodal traffic included in Web totals before the filter launched?
This has been genuinely unclear. Some site owners saw Web numbers dip around the rollout, John Mueller said he believed the traffic wasn’t reported at all before, and trade coverage has reported Google clarifying it was “not previously in the counts.” The safe move: treat pre- and post-rollout Web totals as not directly comparable, and compare text-based to text-based or multimodal to multimodal only.
Is multimodal the same as the Google Images search type?
No. The Images search type tracks the Google Images vertical. Multimodal tracks standard web search results that were triggered by a search using an image — the results are web pages, but the search started visually.
How far back does multimodal data go?
Google hasn’t announced a common historical start date for all properties. Independent launch-week observations found data going back to around September 10, 2026, on some properties. Check your own property’s Dates tab to see the earliest available data.
Sources
- Google Search Central Blog, “Announcing web multimodal Search performance reporting in Search Console,” September 24, 2026 — developers.google.com/search/blog/2026/09/web-multimodal-in-sc
- Search Engine Journal, “Google Search Console Adds Multimodal Filter For Image-Based Searches,” September 24, 2026 — searchenginejournal.com
- Search Engine Roundtable, “Google Search Console Performance Report Gains Multimodal Search Type Filter,” September 24, 2026 — seroundtable.com
- Sunny Patel, “GSC Multimodal Search Filter: What the First Data Shows,” October 2026 — sunnypatel.co.uk
- The Keyword, “Search Console Now Reports Traffic From Image Searches,” September 2026 — thekeyword.co
- BizScore AI, “Search Console Adds Multimodal Filter,” October 2026 — bizscoreai.com
- Carroll Consulting Services, “Search Console Adds a Multimodal Filter as Google Starts a Two-Week Spam Update,” September 28, 2026 — carrollconsultingservices.com
Join the discussion
Have a question or a perspective to share?