AI NewsGo-to-marketAnnouncement
Google adds brand, multimodal and country filters to its Search Analytics API
Google added three filters to the Search Console API: branded versus non-branded queries, web multimodal traffic, and multi-country and multi-device filtering.

Image: Google
Why it mattersSEO teams can now split brand from non-brand and track Lens traffic in one API call. Custom dashboards no longer need keyword regex lists.
An SEO team running a custom dashboard against the Search Console API has done one awkward thing for years. It has kept a regex list of its own brand terms, so it can split brand traffic from non-brand. Google just took that list off the team's hands. Google added three new options to the Search Analytics API, in a single update. The post's authors are the Search Console team: Daniel Waisberg, Nati Yosephian, and Karin Gadasi.
What is new in the API
The first option is a brand filter. The API can now return rows grouped by brandedClassification, or filter them to branded or non-branded only. Google only returns this data for "eligible properties". Other properties see no rows under this field. The second option is a new type value, webMultimodal. It returns traffic from Google Lens, Circle to Search on Android, and image uploads in Google Search. The third option is called filterLists. It lets a single request filter by several countries or several devices at once. The data from webMultimodal is also written to BigQuery for properties that have bulk data export turned on.
What the changes are not
Google did not change what the API measures. Impressions, clicks, CTR and position are the same. Google did not add AI Overviews or AI Mode breakouts to this update. The brand filter is Google's own classification of what counts as branded. It is not a regex the site owner can set. Teams that want their own rule will still need to keep one. The post does not give a date for when AI answer surfaces might be reported in the API.
Why the brand filter matters
An in-house SEO job spends a lot of time on one task: separating "search for the brand name" from "search for the problem the brand solves". The split decides which content to write and which pages to prune. It also decides whether the growth last quarter came from marketing or from search work. A regex list breaks every time the brand team adds a new product line. Google has moved the rule server-side, for properties it thinks have a strong enough brand signal. That is a quiet shift. For sites where Google counts the brand as eligible, the team no longer owns the list.
Why web multimodal matters
A product page that ranks well on an image search has been hidden in the main Performance report. A traveller takes a photo of a product and asks Lens to find it. That click appears somewhere in the aggregate, with no way to pull it out. The new webMultimodal type separates these queries. A retailer, a real estate site or a recipe site now gets its first look at this traffic source. It has been moving while invisible.
What changes for a team
A small SEO team gets to delete code. The custom dashboard that stitches two queries together to get brand-split numbers becomes one request. Teams that already have a BigQuery export can add webMultimodal alongside their existing tables. Teams running dashboards per region can switch from one API call per country to one call across several. That cuts quota use and request count.
The brand filter has one catch worth saying out loud. Eligibility is Google's call, and Google has not published the rule in detail. A site that is not eligible gets no rows at all, not a different answer. A team that depends on the filter needs to check whether the property returns data. The old regex list stays in place for the properties that do not.
Source
- New filtering and reporting capabilities in the Search Analytics API, Daniel Waisberg, Nati Yosephian and Karin Gadasi, Google Search Central.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.


