Share:

AI marketing & automation




Google added three Search Console Performance report features to the Search Analytics API on November 16, 2026. You now get a native brand split, image-search data and multi-country filters from one API. Google’s announcement says the change came “following ecosystem requests to make this data available for custom reporting and automated workflows”.
Our view: use the branded split for trend and discovery reporting now, but keep your regex list as the system of record for client-facing brand numbers until you reconcile the two. This article covers what changed, where each option breaks, and a five-part readiness test you can run on every property this week.
Google added three options on November 16, 2026. Group or filter by brandedClassification to split branded from non-branded queries. Set type to webMultimodal for Lens, Circle to Search and image upload data. Use filterLists to match several countries or devices in one request. Each option has its own limits.
The Search Analytics API reference lists the new fields and values. This table shows the catch for each one.
| Option | API field | Main catch |
|---|---|---|
| Branded split | brandedClassification with BRANDED or NON_BRANDED | Eligible properties only, and no query filter in the same request |
| Multimodal data | type set to webMultimodal | Covers Lens, Circle to Search on Android and image uploads |
| Several markets or devices | filterLists | Accepts only country or device |
rowLimit runs from 1 to 25,000 with a default of 1,000. The API returns top rows, not necessarily all rows. groupType supports only “and”.
Not yet for client-facing numbers. Google’s classifier is AI-assisted, not keyword matching. It returns data only for eligible properties, and its history starts February 21, 2026. Use it now for trend and discovery reporting. Keep your regex list as the system of record until you reconcile the two methods on your own properties.
Google’s branded queries filter post says the system recognizes your brand name in all languages, including typos, and flags queries that refer to a unique product or service of your site. A regex list only catches the terms you typed into it. Google also says some queries may be misidentified.
You cannot tune the classifier. In Search Engine Journal’s Q&A, Google’s John Mueller said, “At the moment, I’m not aware of plans to provide customization.”
Regex splits miss clicks too. A 2022 Ahrefs study found anonymous queries accounted for 46.08 percent of all clicks, across 146,741 websites and nearly 9 billion clicks in one month of data. A regex only sees queries that Search Console shows, so check how much of your own click total has no visible query.
In a client account analysis of 40 properties, Launchcodex ran a brandedClassification request, and 24 returned rows. The other 16 came back empty. The branded filter is not available for URL path or subdomain properties, and it needs enough query and impression volume.
We then compared both methods over the same 90 day window on the 24 eligible properties. The API branded click total differed from our regex brand total by a median of 12 percent, and by more than 25 percent on 5 properties.
“Report non-branded growth as your SEO line and show branded clicks beside it. A strong brand month can hide a weak discovery month.” Tanner Medina, Co-Founder & Chief Growth Officer
History is short too. As of November 16, 2026, branded data covers 268 days, so year-over-year brand comparisons are not possible through the API yet.
Add a filterLists entry with the country dimension and list 3-letter ISO 3166-1 alpha-3 codes. A row matches if its country equals any code in the list. Google’s own examples are groups such as EMEA, LATAM and DACH. Add a device entry the same way to cover several device types.
Google’s post says the old route was combining separate requests. In an audit of 18 multi-market client reporting setups, 13 pulled each country separately and added the results in a spreadsheet.
Say an agency reports on 30 properties across six markets each. Six calls per property means 180 calls per run. One filterLists call per property cuts that to 30.
Quota is not the main gain. Search Console’s usage limits allow 1,200 queries per minute per site for Search Analytics, so six calls per property fit easily. The gain is one request to maintain and no spreadsheet step where a combined total can break. The Performance report added multi-country filtering about six weeks earlier, on October 6, 2026, according to Search Engine Roundtable.
It shows how image-based searches reach your pages. Set type to webMultimodal to get performance data from Google Lens, Circle to Search on Android and image uploads in Google Search. Google says metrics appear only if your site receives this traffic. Pull it as its own series until you confirm how it relates to your web totals.
Multimodal searches include an image in the query. Google said in February 2025 that Lens is used for more than 20 billion visual searches every month. Google’s product leads for Search Console and Lens, Moshe Samet and Harsh Kharbanda, described the reporting in Search Engine Journal this way: “This update is designed to give you insights into how your content is surfaced”.
In the same account analysis, we checked all 40 client properties for webMultimodal data, and 8 returned any rows.
“With 8 of 40 properties returning rows, I would treat image search as a watch metric, not a target.” Tanner Medina, Co-Founder & Chief Growth Officer
Check coverage too. Google’s web multimodal announcement lists the Chrome “Search this image” option among the surfaces, while the API reference lists three. Treat Chrome coverage in the API as unconfirmed.
The readiness test is five checks you run on each property before you rebuild a report on the new options: eligibility, reconciliation, scope, limits and multimodal baseline. Each check catches one failure, such as a blank dashboard or a changed brand history. Run them in this order.
“I put both series in one Google Sheets tab and let the gap run for a full month before a Looker Studio chart switches over.” Derick Do, Co-Founder & Chief Product Officer
Start with one property and the eligibility test. Use the API branded split for trend reporting, keep your regex list for client-facing brand numbers, and hold multimodal data as its own series until you have checked how it relates to your web totals.
Pick your highest-traffic property and run the eligibility test today. Launchcodex builds reporting pipelines like these as part of its data infrastructure work, and can help you run the test across your properties.
Yes. Group or filter by brandedClassification with BRANDED or NON_BRANDED. Data starts February 21, 2026, and covers only eligible properties. Other properties return no rows.
No. Google’s reference bars brandedClassification filtering or grouping from sharing a request with a query filter.
Use 3-letter ISO 3166-1 alpha-3 codes. Device values are DESKTOP, MOBILE and TABLET.
Google says it is, for properties with an active bulk data export. The bulk export guide says setup needs a Google Cloud project with billing, the first export runs up to 48 hours later, and data from before setup is excluded.



Real stories from the people we’ve partnered with to modernize and grow their marketing.