To find image-based search traffic in Google Search Console, open Performance → Search results, select Search type, and choose Web: multimodal. This view reports web results from searches that included an image, such as a photo taken with Google Lens or a screenshot searched with Circle to Search. Review the Pages tab to see which URLs appeared, and use the chart to check clicks and impressions. Google began rolling out the filter globally on September 24, 2026, so a property may not yet have data to show. Google's announcement explains the rollout.
The new filter answers a specific measurement question: which pages receive visibility or visits from image-led web searches? It does not identify the exact photo someone used, separate Lens from every other image input, or tell you that a page needs an SEO fix. The steps below show how to read the report before deciding what to inspect on your site.
What does “Web: multimodal” count?
Google defines Web: multimodal as web search results shown when an image was used as part of the search. Its launch announcement names Google Lens, Circle to Search on Android, images uploaded to Google Search, and Chrome's right-click Search this image action. These are ways to start a search, not separate rows you can select in this filter.
The distinction between Web: multimodal and Image in Search Console matters:
| Search type | What it groups | Question it helps answer |
|---|---|---|
| Web: multimodal | Web results from searches that used an image as input | Which of my pages surfaced after an image-led search? |
| Image | Results in Google's image-search experience | How did my images perform as image results? |
Google's Performance report documentation lists them as different search types. A product page may have data in one, both, or neither. Do not treat the two figures as the same measure or as a complete count of visual discovery. Start with Web: multimodal when you want to investigate image-led web discovery; use the Image search type for the separate question of image-result performance.
Google's announcement describes a reporting change, not a new ranking factor or a requirement to add “multimodal SEO” markup. You can use the data to choose a page for closer inspection, but the filter alone cannot explain why Google showed that page.
How to find image-based search traffic in Search Console
Use Google's own Search Console interface for this report. A connected SEO tool may import other Search Console data without exposing this newly added search-type filter.
- Choose the correct property. Open Search Console and select the domain or URL-prefix property for the site you want to measure. If you manage both, keep the choice consistent when comparing periods; their coverage can differ.
- Open Performance → Search results. At the top of the report, select Search type, then Web: multimodal. Google's report help distinguishes this from Web: text-based, Image, Video, and News.
- Choose a relevant date range. The rollout began on September 24, 2026. Avoid assuming every property has the same amount of historical multimodal data. Start with an available period that includes the rollout, then note the first date your property actually reports.
- Open the Pages tab. Sort by impressions to find URLs shown most often in image-led web searches; sort by clicks to find URLs bringing visits. Use Devices and Countries to see where the reported activity occurred. A page filter lets you narrow the view to one URL.
- Export the view if you need a baseline. Search Console's Export control downloads the chart and table data. Record the property, date range, search type, and any page filter alongside the file so a later comparison uses the same scope.
You should now have a list of reported pages, not a list of the photographs people searched with. Look for a page whose multimodal visibility matters to your business before changing any images. A large impression count with few clicks is a reason to investigate the page and result context, not proof that its alt text, image quality, or ranking is defective.
How to read the numbers for one page
The Search results report can show clicks, impressions, CTR, and average position. A click means someone selected a result leading to your site; an impression means your site appeared in a result. CTR is clicks divided by impressions. Average position summarizes positions across reported results, so it is not a fixed ranking for one picture or one query. Google's Performance report help also notes that chart totals and page-table totals can differ because they use different aggregation levels.
Consider this illustrative example, not Screpy data or a measured result. Select one product URL and the same available date range under each web search type:
| Selected page | Search type | Impressions | Clicks | CTR |
|---|---|---|---|---|
/products/blue-ceramic-mug/ |
Web: multimodal | 240 | 6 | 2.5% |
/products/blue-ceramic-mug/ |
Web: text-based | 3,400 | 102 | 3.0% |
The page appeared in both kinds of web search. The multimodal row tells you that image-led searches surfaced this page and produced some visits in the selected period. It does not tell you which image the searcher supplied, whether the result came from Lens or Circle to Search, or why someone did not click. The smaller multimodal sample also makes a 0.5-point CTR difference weak evidence for an optimization decision.
Your next question is about the page, not the percentage: does the landing page clearly show the pictured product, its distinguishing features, and accurate supporting information? If the page is valuable and its images are broken, hidden, misleading, or missing useful context, those are concrete issues to check. If the page looks sound, keep the row as a baseline and look for a sustained pattern before assigning a cause.
What does the Generative AI report add?
Google also added Web: multimodal to Performance → Generative AI. Apply the same Search type there when you want to see whether pages were shown in Google Search's AI Overviews or AI Mode during image-led web searches. The Generative AI report documentation says this report measures organic impressions for those features and can group them by page, country, device, and date.
| Open this report | To answer this question | Main metric |
|---|---|---|
| Performance → Search results | Did an image-led web search show or send a visitor to my page? | Clicks and impressions, with CTR and average position |
| Performance → Generative AI | Was my page linked in a supported AI feature for an image-led web search? | Impressions |
Keep these reports separate when you present results. An AI impression is not a click, lead, sale, or proof that the AI feature caused a later visit. The report groups supported generative features in its published scope; do not infer a separate Lens-to-AI-Overview conversion funnel from these numbers. If you need to investigate a page, use the report's Pages dimension to choose the URL, then inspect that page and its search context rather than treating an impression as a quality score.
Why might the multimodal view show no data?
An empty chart is not a diagnosis. Google began a global rollout on September 24 and says the new metrics appear when a site receives traffic from qualifying searches. Check what you selected before concluding that an image or page has failed SEO.
| What you see | Check first | What you can conclude |
|---|---|---|
| The filter is absent or unavailable | Confirm the property, account access, and whether the rollout has reached the view | You cannot yet diagnose image performance from that missing view |
| The filter opens but shows no rows | Select a date range that includes available data; remove restrictive page, country, or device filters | There may be no reportable activity in that scope; absence alone is not an indexing verdict |
The exported file has zeros where the UI showed ~ or - |
Recheck the original interface before using the CSV in a calculation | Google says unavailable values export as zero, which is not necessarily a measured zero |
The Search results report help documents the export behavior and explains why chart totals and table totals can differ. If you are investigating a particular image, remember that this is a combined web-search category. It cannot reveal the uploaded photograph or give you a separate count for Lens versus the other image-led entry points.
What should you check after finding a page?
Use the multimodal report to choose what to inspect, then verify the page itself. A reporting row is evidence of appearance or clicks in the selected search type; it is not a list of failed image-SEO checks.
- Pick a page that matters. Start with a product, category, guide, or other URL whose visual subject is central to the reader's task. Compare its multimodal data with the same page's text-based data over the same dates. Record the current numbers before making changes, especially while the new report has only a short history.
- Inspect the page and image, not just its CTR. Is the main image accessible in an HTML
<img>element? Is it relevant to the page, sharp enough for visitors, and supported by nearby text that names the product or subject accurately? Google's image SEO guidance says it can find images in thesrcattribute of HTML image elements but does not index CSS background images. It also explains how page context and useful alt text help Google understand an image. If the mug in the example is only a CSS background, that is a concrete technical issue to fix; the report did not prove that this specific image triggered the impressions. - Check whether the problem repeats. Screpy's Website Audit can group crawled images by response health, source type, alt text, dimensions, responsive delivery, and the pages that use them. Its alt-text audit guide helps distinguish informative product images from decorative images. Fix the affected page or shared template, recrawl to verify the technical change, then watch comparable Search Console periods for a sustained change. A better crawl result confirms the site fix, not a guaranteed rise in multimodal clicks.
If you already connect Search Console to Screpy, keep the tools' jobs clear: the Screpy integration guide covers imported search performance and page investigation, while Google's native Search Console is where this newly announced Web: multimodal filter is documented. That distinction prevents a reader from searching for a control Screpy has not documented as available.