Your camera traffic has pages, not queries

Web multimodal is the Search Console search type for Google results triggered by an image, photo, or screenshot instead of typed text, and since September 24, 2026 it has its own line in your reports.

The change is small in the interface and large in what it exposes. For years, a person pointing a phone at a product, a plant, or a broken part produced a Google search that landed on someone's page, and that page owner had no clean way to see it. Now the traffic is visible. What it does not come with is the thing SEO reporting is built around: the query.

#What Google shipped

On September 24, 2026, Google announced web multimodal reporting in Search Console. The Web search type in the Performance report now splits into two options. Google's Performance report help page defines them this way: text-based covers "search traffic originating from text queries entered into the standard Google search bar," and multimodal covers searches where people use images, including Google Lens, Circle to Search on Android, uploaded images, and Chrome image search.

The filter appears in two places, according to Search Engine Roundtable's coverage: the classic Performance report for Search results and the separate generative AI features report. That second report is the one Google introduced in June 2026 for impressions in AI Overviews, AI Mode and AI features in Discover. So an image-led search that produced an AI answer citing your page now has somewhere to show up.

Two details change how to read the numbers.

First, this is new data, not a relabel. John Mueller said "the data wasn't previously in the counts," as reported by Search Engine Roundtable. Image-led traffic was not hiding inside your Web totals. It was missing from them.

Second, the Queries dimension is not available. Google's explanation, quoted in Search Engine Journal, is that "because multimodal searches mostly use images rather than text, specific text query data isn't available for this traffic." You still get pages, countries, devices, dates, clicks, impressions, CTR and position.

#Why the volume is worth a separate report

Google has published one usable scale figure for camera search. In October 2024 it said people use Lens for "nearly 20 billion visual searches" every month, and that 20 percent of Lens searches are shopping related. That number is two years old and covers Lens alone, so treat it as a floor on the category rather than a current measurement. It is still large enough that a retailer, a parts supplier, a recipe site or a travel publisher should expect at least some pages with real multimodal impressions.

The surfaces matter too. Circle to Search and Chrome's "Search this image" are not apps someone opens to search. They are actions taken on top of something already on screen: a social post, a video frame, a screenshot of a competitor's product. That makes this the closest Search Console has come to reporting demand that started outside Google.

#The page becomes the unit of analysis

Without queries, every multimodal question has to be answered at the page level. That is a different discipline from text SEO, where you start from the query and work back to the page.

For a page with multimodal impressions, the useful question is which visual on that page made it a match. A product page might win because its main photo shows the item from the angle people photograph it in a store. A how-to page might win because one diagram matches the part someone is holding. A local page might win on a storefront photo. Search Console will not tell you which image did the work, so the inference is yours.

A practical way to approach it:

1. Sort multimodal pages by impressions and group them by page type. Product, recipe, plant or species, landmark, repair or parts, and so on. The pattern by type tells you more than any single page.
2. Inventory the images on the top pages. Note which are original photography and which are stock or manufacturer images shared with every other seller. If the same image appears on every reseller's site, matching it does not favor you.
3. Compare against text-based performance for the same pages. A page that does well in multimodal and poorly in text is a page whose visuals are ahead of its copy, and the reverse is common for thin catalog pages.
4. Check the generative AI features report with the multimodal filter applied. Pages cited in image-led AI answers are worth reading closely, because the answer had to connect a picture to your text.

Google's own image SEO guidance still describes the inputs you control: descriptive filenames, alt text, useful surrounding text, and placing important images near the content they illustrate. Nothing in the announcement suggests a separate set of rules for multimodal ranking, so those basics are the starting point.

#The API does not have the split yet

If your reporting runs through the Search Analytics API into a dashboard or warehouse, you cannot pull multimodal separately today. Search Engine Journal noted that the API reference, last updated August 11, lists web, image, video, news, Discover and Google News as report types, with no multimodal value. Google has not documented whether API web rows now include image-led searches.

Until that changes, the reliable path is the Export button in the Performance report with the multimodal filter set. That is a manual step, so schedule it. A weekly export by page keeps enough history to see trends; an occasional export does not.

#What to do this week

• Annotate September 24, 2026 in every dashboard that reads Search Console. If a property shows multimodal data starting there, any comparison that crosses that date needs care.
• Keep text-based and multimodal CTR apart. Someone who photographed a product and someone who typed a product name are at different points in the decision. Blending the two will make both look worse than they are.
• Export multimodal by page on a schedule until the API catches up, and store it next to your text-based data.
• Replace shared images on high-value pages. Original photography of the product, the place, or the result is the one image signal a competitor cannot copy.
• Treat position loosely. Google has not explained how position is assigned for a result reached from an image, so compare multimodal positions only with themselves over time.

The report does not answer what people photographed. It does answer, for the first time, which of your pages a camera found. Start from that list.