Google Search Console now separates web performance data into two buckets: Web: text-based and Web: multimodal. The multimodal bucket counts visits triggered when a searcher uses an image, photo, or screenshot instead of typed text, through Google Lens, Circle to Search on Android, image uploads, or the Chrome right-click “Search this image” tool. For sites that lean on product photography, recipe imagery, or visual how-to content, this is the first time that traffic has been isolated in a reportable metric at all.
What makes the rollout unusual is the accounting behind it. Google’s own Search Console help documentation now states plainly that “Query data for multimodal searches” is not available “because multimodal searches mostly use images rather than text,” so the queries dimension disappears entirely once a site owner selects the multimodal filter. A property can see impressions and clicks tied to image-based discovery, but not the query string that produced them, because in most cases there was never a text query to log.
Google Search Advocate John Mueller addressed the more consequential question directly on Bluesky: was this multimodal traffic already folded into the existing “web” totals, meaning today’s split just relabels old numbers, or is it genuinely new? Asked by user Dave Smart whether multimodal impressions had previously been “grouped under ‘web’ before,” Mueller first replied, “I’ll double-check, I thought these were not reported at all before.” He followed up after checking internally: “And … checking with the team, the data wasn’t previously in the counts, the report has new data.” That confirmation, reported first by Search Engine Roundtable, rules out the simpler explanation that this is a re-slice of existing web totals; it is additional visibility into activity Search Console was not counting before.
Mueller also framed the release himself, writing that “If you have an image-heavy website, Search Console now reports on ‘multimodal’ queries like when you take a photo w/Lens,” while noting the underlying image itself stays hidden from the report. Google’s official announcement, published Wednesday, confirmed the filter is rolling out globally in both the Performance report for search results and the report for Generative AI features, and that data appears once a property starts receiving qualifying traffic.
The missing queries dimension changes how reporting on this traffic has to work. Analysts accustomed to pivoting performance data by search term will instead need to lean on page-level and date-level breakdowns to spot which URLs benefit from image discovery, since there is no keyword layer to filter by. That also shifts the optimization target: without a query string to chase, the signal an operator can actually influence is the image itself, meaning file naming, alt text, structured data, and photo quality become the levers for multimodal visibility rather than the phrasing on the page.
Barry Schwartz noted in his original report that this still does not extend to click or query data inside the generative AI performance report, so the gap in reporting granularity persists across Google’s newer surfaces, not just this one filter.
Site owners with substantial image inventory should pull the new Web: multimodal segment now and benchmark its impression volume against total web impressions before drawing conclusions from any month-over-month comparison, since the “new data” framing means there is no historical baseline before this rollout to compare it against.
Reported by Barry Schwartz at Search Engine Roundtable on September 24, 2026.
