Skip to content
Performance Marketing Hub
SEO

Search Console Branded Queries: How to Split Brand and Non-Brand Traffic in the Report and the API

Search Console branded queries are sorted by an AI assisted classifier into Branded and Non-branded groups, in the Performance report and, since November 16, 2026, in the Search Analytics API through the brandedClassification dimension. The data covers eligible top level properties from February 21, 2026, so older history still needs a regex brand filter.

Maxim Baeten
By Freelance digital performance marketer
Updated 8 min read

Key takeaways

  • Search Console branded queries are classified by an internal AI assisted system, not by a regex, and include the brand name in all languages, its misspellings and queries about unique products or services of the site.
  • The Search Analytics API added the brandedClassification dimension on November 16, 2026, with the values BRANDED and NON_BRANDED, the equals operator only and no combination with a query filter in the same request.
  • Branded data exists only for top level properties with enough queries and impressions, starts on February 21, 2026 according to the API reference, and comes back as an empty response for ineligible properties.
  • A custom regex brand filter is still needed for dates before February 21, 2026, for URL path and subdomain properties, for low volume sites and for any request that also filters on query text.
  • Google describes non-branded queries as the ones that show how new users find a site, so non-brand clicks are the cleaner metric for judging SEO growth and the impact of AI Overviews or a core update.

Search Console branded queries are searches that Google's AI assisted classifier ties to your brand. You split them from non-branded queries with the Branded and Non-branded Query filter options or, since November 16, 2026, with brandedClassification, the Search Analytics API dimension that labels each query as BRANDED or NON_BRANDED. Data starts February 21, 2026, for eligible properties.

This guide is part of the SEO hub and assumes you know the basics of the Performance report from how to use Google Search Console. It covers the report filter, the Insights card, the API request and the cases where a custom regex brand filter still does a better job. A custom regex brand filter is a Custom (regex) rule in the Query filter that matches a brand term list you maintain yourself. The facts reflect Google's documentation as of November 2026.

How does Search Console decide which queries are branded?

Search Console decides which queries are branded with an internal, AI assisted system, not with a keyword list or a regular expression. A branded query contains the site's brand name in any language, a misspelling or variation of it, or a unique product or service of the brand, even when the brand name itself is missing.

Google's November 2025 announcement of the branded queries filter uses itself as the example: "Gogle" counts as a misspelling and "Gmail" counts as a brand product of google.com. Non-branded is every other query. The Performance report help page defines branded queries as searches that include "your brand name, domain, or a brand-specific product or service."

The classification is contextual, so it is not exact. Google warns that "some queries may occasionally be misidentified," and the Insights report help page says query classification "is powered by AI and may mislabel" queries. You cannot correct it yourself. Asked whether site owners could add terms, John Mueller answered "Not at the moment," as Search Engine Journal reported on March 12, 2026.

The split is a reporting view only. Google states the filter "has no effect on how Google Search ranking works."

Why do some properties get no Search Console branded queries data?

Some properties get no Search Console branded queries data because the classifier only covers top level properties with enough queries and impressions. URL path properties, subdomain properties and small sites see no Branded option in the report, and the Search Analytics API sends them a response without any rows.

Google rolled the filter out gradually from November 20, 2025 and, per an update to Google's branded queries filter announcement, made it available to all eligible sites on March 11, 2026. A Domain property such as sc-domain:example.com qualifies when it has the volume; a folder property such as https://example.com/blog/ never does.

Time is the second limit. The Search Analytics API reference states that branded data "is available starting February 21, 2026." Mueller confirmed in March 2026, as Search Engine Journal reported, that the classification is not applied backward: "there's a point when the data starts being tracked, and you'll see that in the report if you look far enough back." The Performance report help page words the start date differently, so check the first day with branded data in your own property before you build a trend. Until February 21, 2027, no property has a full year of branded history.

How do you split brand and non-brand traffic in the Performance report?

You split brand and non-brand traffic in the Performance report by opening the Query filter and choosing Branded or Non-branded. Clicks, impressions, CTR and average position then cover only that group, for the web, image, video and news search types, and the Insights report adds a card with the click split.

  1. Open the right property. Use a Domain property or a top level URL prefix property, then open Performance and Search results.
  2. Set the date range. Start on or after the first day with branded data in your property.
  3. Add the Query filter. Click the filter bar, select the Query filter and choose Branded or Non-branded, as the help page describes.
  4. Read all four metrics. Turn on clicks, impressions, CTR and average position for the selected group.
  5. Export both groups. Repeat with the other option over the same dates and export both views for your report.

The Insights report shows the same split as a card called "Branded and non-branded traffic," which compares clicks from people who searched for the brand with clicks from general searches. Use the card for a quick read and the filter when you need pages, countries or devices per group.

How do you request brandedClassification in the Search Analytics API?

You request brand data in the Search Analytics API by adding brandedClassification as a dimension to group by, or as a filter with the equals operator and the value BRANDED or NON_BRANDED. Google added it on November 16, 2026, together with the webMultimodal search type and a new filterLists field.

Google's API announcement says the change lets you "automatically segment branded from non-branded traffic in your dashboards" instead of "maintaining complex keyword regex lists." The API reference sets three rules:

  • Equals only. brandedClassification accepts no contains, notEquals or regex operator.
  • No query filter in the same request. Filtering or grouping by brandedClassification "can't be combined with a "query" filter in the same request."
  • Eligible properties only. Other properties get no rows, not an error message.

This request body, sent with POST to the searchAnalytics/query endpoint of a property, returns daily non-brand clicks for Germany, Austria and Switzerland:

{
  "startDate": "2026-10-01",
  "endDate": "2026-10-31",
  "dimensions": ["date"],
  "type": "web",
  "dimensionFilterGroups": [{
    "filters": [
      { "dimension": "brandedClassification", "operator": "equals", "expression": "NON_BRANDED" }
    ],
    "filterLists": [
      { "dimension": "country", "expressions": ["deu", "aut", "che"] }
    ]
  }]
}

The filterLists field matches one dimension against several exact values and accepts only country (ISO 3166-1 alpha-3 codes) and device. A request cannot filter the same dimension in both filters and filterLists. To get both groups in one call, drop the filter and add brandedClassification to the dimensions. Lens and Circle to Search traffic has its own webMultimodal type, covered in Search Console multimodal search.

When does a custom brand regex still beat the built-in classifier?

A custom brand regex still beats the built-in classifier when you need dates before February 21, 2026, when the property is a URL path, subdomain or low volume site, when you must control exactly which terms count as brand, or when the same request filters on query text. In every other case, start with the classifier.

The Brand Split Decision Table matches each reporting situation to a brand method. It is the author's rule, not Google guidance, and every limit in it comes from the Google documentation cited in this guide.

SituationUseWhy
Eligible property, dates from February 21, 2026Built-in filter or API dimensionGoogle maintains typos, languages and product names
Comparison that reaches before February 21, 2026Regex on both periodsThe classifier has no older data, and mixed methods break the trend
URL path, subdomain or low volume propertyRegexNo branded data is returned
Brand split plus a topic, such as non-brand queries with "pricing"Two API calls, or regex in one requestbrandedClassification cannot be combined with a query filter
Sub-brand or product line owned by another teamRegexThe classifier cannot be edited
Audit of the classifier itselfBoth, side by sideDisagreements expose misclassified queries
Brand Split Decision Table: which brand method to use per situation (author's rule, not Google guidance).

Regex filters have their own cost. The advanced filtering help page says query filters "can affect the report totals due to data truncation and the omission of anonymized queries," so a regex brand view plus a regex non-brand view will not add up to the unfiltered total. Search Console regex uses RE2 syntax and is not case sensitive by default. Search Engine Journal advises running the API split next to an existing brand regex and comparing where they disagree.

How do you report non-brand growth after AI Overviews and core updates?

You report non-brand growth by tracking non-branded clicks and impressions as their own trend line, next to branded clicks, instead of one blended total. Google describes non-branded queries as the ones that show how new users find a site, so they carry the signal for SEO work, AI Overviews and core update effects.

Blended numbers hide that signal. According to Google's announcement, branded queries "typically lead to higher-ranking pages from your site and result in higher click-through rates." When the brand share of traffic shifts, blended CTR and position shift with it, even if no ranking changed.

A hypothetical worked example shows the effect. In the hypothetical case, a site gets 6,000 branded and 4,000 non-branded clicks in one month. In the hypothetical next month, a brand campaign lifts branded clicks to 7,500 while non-branded clicks fall to 3,200. Total clicks grow 7 percent, and the 20 percent loss in non-brand clicks stays invisible in the blended total.

AI Overviews and core updates change how pages rank and get clicked for topic queries, so read their effect on the non-branded line first; a brand campaign or press coverage moves the branded line. For the measurement side, see AI Overviews click through rate and what to do during a Google core update.

Common mistakes with Search Console branded queries

The common mistakes with Search Console branded queries come from treating the classifier as a complete, exact and editable brand list. It is none of those, and each mistake below leads to a wrong trend, an empty dashboard or a request that cannot run, with a fix that needs only the documentation and a regex.

  • Treating the classifier as exact. Google says queries can be misidentified. Spot check the top branded and non-branded queries every quarter.
  • Comparing across February 21, 2026. Older periods have no branded data, so a non-brand trend that starts there looks like growth from zero. Use a regex for both periods.
  • Reading an empty API response as zero traffic. Ineligible properties get no rows. Flag empty responses in the dashboard instead of plotting them as zero.
  • Adding a query filter to a brand request. The API does not allow it. Run the brand call and the topic call separately and join the rows.
  • Using a folder property. Folder and subdomain properties never get the split. Switch to the Domain property and filter pages there.
  • Reporting blended CTR as an SEO result. A higher brand share raises CTR without any ranking change. Report CTR per group.

FAQ

Why don't I see the branded queries filter in my Search Console?

The filter only appears for top level properties with enough queries and impressions. URL path properties such as a blog folder, subdomain properties and small sites do not get it. Open a Domain property for the same site and check again.

Can I add my own brand terms to the branded queries filter?

No, Google does not let site owners edit the list. John Mueller said in March 2026 that he was not aware of plans for customization and pointed to the feedback buttons in the tool. Use a custom regex in the Query filter when you need full control over which terms count as brand.

Do misspelled brand searches count as branded queries in Search Console?

Yes, Google says the classifier includes typos and variations of the brand name, the brand name in all languages and searches for unique products or services of the brand. Because the system is AI assisted, some queries can still be misidentified. Spot check the top queries in both groups.

Can I see branded and non-branded traffic from before 2026?

No, the classification is not applied retroactively. The API reference says branded data starts on February 21, 2026. For older periods, build a regex brand filter and apply it to both periods you compare.

Does the branded filter change how my site ranks in Google?

No, Google states the branded queries filter has no effect on how Google Search ranking works. It only changes how Search Console groups the data you already have. Your rankings and clicks stay the same whichever view you pick.

Sources and further reading

Maxim Baeten
Written byMaxim Baeten

Digital performance marketer from Belgium, specialised in lead generation for B2B (SaaS) software companies. Full-time freelancer since January 2019.