In short and in plain language
Search Console is a tool from Google. It shows how people find a website on Google. Some people search for the name of the firm. Others search for a topic or a service.
Google now splits these two groups by itself. There is a filter for this. You pick Branded or Non-branded. A Google program sorts the searches. Now and then it makes mistakes.
Not every website has the filter. The website needs enough searches. And the whole site must be in Search Console, not just one part of it.
For work on search, the second group counts most. These people did not know the firm before. When their part grows, the work on the website pays off.
At the end of the year, it helps to look back at the year before. You can write key events as notes on the chart. Then you still understand the numbers later.
Anyone who looks at the clicks from Google Search in Search Console sees the sum of two very different groups. One group already knows the business and types its name, perhaps after a recommendation, a flyer or an ad. The other is looking for a service, a product or an answer and comes across the website along the way. When the question is whether search engine optimization works, the second group is the one that matters most. Until now, website owners had to strip out branded searches by hand. On November 20, 2025, Google announced a dedicated filter in the performance report that automatically separates branded from non-branded queries (Google Search Central Blog); since March 11, 2026, it has been available to all eligible sites (Google Search Central Blog). This article explains what the filter counts, when it is missing, where its classification reaches its limits, how the share of non-branded clicks becomes a measure of SEO work, how a custom regex filter serves as a cross-check and how to run the year-over-year comparison at the end of the year.
Key takeaways
- Since March 11, 2026, the Search Console performance report separates branded and non-branded queries for all eligible sites (Google Search Central Blog). The classification comes from an AI-assisted system, not from a regex filter.
- The filter only appears for top-level properties with enough queries and impressions, not for URL path or subdomain properties. If it is missing, check the property type first and add a Domain property.
- Some queries may be misclassified, and anonymized queries drop out when a query filter is applied. The table shows at most 1,000 rows (Google Search Console); a monthly spot check reveals misclassifications.
- The share of non-branded clicks shows how many visitors find the website through topics rather than its name. Because half (50 percent) of internet users aged 16 and over at least sometimes use AI chats instead of classic search (Bitkom), impressions belong in the picture.
- A custom regex filter in RE2 syntax serves as a cross-check on the automatic classification and works in every property. Through the Search Console API it returns up to 25,000 rows per request (Google Search Console API).
- For the year-over-year comparison, show the last 16 months and compare the last 3 months with the previous year (Google Search Central). Annotations of up to 120 characters (Google Search Console) record relaunches, campaigns and renamings in the chart.
What the branded filter counts
Google describes a branded query as a query that includes the brand name, variations or misspellings of that name, and products and services that belong to the brand. The announcement uses the name Google itself, the misspelling Gogle and the product Gmail as examples (Google Search Central Blog). Applied to a small or mid-sized business, that means the company name, the company name together with a city or legal form, common misspellings and the name of an in-house product that only this provider sells. The performance report help explicitly includes the site's own domain as well.
In the performance report for search results, the filter splits query data into two views (Google Search Central Blog). In the English interface they are called Branded and Non-branded; you reach them through the filter bar and the query filter. The filter works in all four search types, meaning web, image, video and news (Google Search Central Blog), and shows clicks, impressions, average position and click-through rate for the selected group only. Google has also added a card to the Insights report that breaks down all clicks into branded and non-branded traffic.
What matters is how the classification is made. It is not based on a regular expression that includes or excludes terms but on an internal, AI-assisted system. It recognizes the site's brand name in all languages, plus misspellings and even queries without the brand name that refer to a unique product or service of the site. According to Google, the filter has no effect on ranking; it only serves to segment and analyze the data in Search Console.
Name and domain
The company name, also with a city, legal form or industry term before or after it, plus the site's own domain as a search term.
Variations and misspellings
Typos, missing or swapped letters and spellings in other languages that the system attributes to the name.
Your own products
Products and services that clearly belong to the website, even if the company name does not appear in the query at all.
The filter is part of a series of analysis aids that Google added to Search Console in quick succession. Query groups in the Insights report, which use AI to bundle similar queries, arrived on October 27, 2025 (Google Search Central Blog), and custom annotations in the performance chart followed on November 17, 2025 (Google Search Central Blog). Together they allow an analysis that used to take spreadsheets and a lot of manual work: grouping queries, separating branded searches and marking events over time.
When the filter is missing: property and data volume
Not every Search Console property shows the filter. Google gives the following reasons: it is only available for top-level properties, not for URL path properties such as example.com/path and not for subdomain properties. The site also needs a sufficient volume of queries and impressions for the classification signals to work. The Insights report help adds that the branded and non-branded traffic card may not appear for all eligible sites until it is fully launched.
Which property type you have is decided when the property is created. A Domain property covers all subdomains such as www or m and all protocols; it is verified through a DNS record. A URL-prefix property, by contrast, only contains addresses that start with the specified prefix including the protocol, and it can be verified in several ways, for example with an HTML file. If you once set up your website as a URL-prefix property because you had no access to DNS management, you may still be working with that variant today.
So if the filter is missing, look at the property first. If your existing property does not show it, an additional Domain property for the main domain can help: it brings all variants of the address together under one roof. The previous property remains in place alongside it. If the search volume is too low, a custom regex filter remains the way to strip out branded searches. If your Search Console is maintained as part of a website care plan, have the property type checked there once.
Quick check in the interface
Where the classification reaches its limits
Google itself points out that some queries may occasionally be misidentified because of the dynamic and contextual nature of brand classification. In the day-to-day reality of small and mid-sized businesses, there are obvious triggers. If the company name consists of a generic term and a surname, a general query about the service may pass as branded or vice versa. If the company has changed its name, some customers keep searching for the old one. And a product name that is also an everyday word can hardly be attributed to a brand without context.
On top of that come limits that apply to every query filter. Search Console does not show very rare queries in order to protect user privacy; these anonymized queries are included in the chart totals as long as no query filter is applied. With such a filter, the totals for matching and non-matching queries therefore do not have to add up to the unfiltered total. Because the branded filter also works on queries, the sum of branded and non-branded can fall below the total. The table below the chart also shows at most 1,000 rows (Google Search Console); rare queries are missing there even though the chart counts them.
Time lag matters for ongoing analysis as well. Collected data is usually available after 2 to 3 days (Google Search Console), and the most recent values can be preliminary and still change. The report records daily values in Pacific Time, and weeks run from Sunday to Saturday. If you compare monthly values with your own incoming inquiries or your point-of-sale system, expect small shifts at the edges of a period.
- Review the queries with the most clicks in the Non-branded view: do they include the company name, an old name, an abbreviation or one of your own products?
- In the Branded view, look for queries that only name a service and a location.
- Note the names of owners, locations and your own products; they are candidates for your custom regex filter.
- Record conspicuous queries with a date so you can see in the following month whether the classification stays stable.
The non-branded share as a measure of SEO work
A website alone hardly sets a business apart anymore: 91.7 percent of enterprises in Germany with at least ten employees had one in 2025 (Eurostat). The difference is whether people who do not yet know the name can find it. Branded and non-branded queries behave differently. According to Google, branded queries typically lead to higher-ranking pages and higher click-through rates, whereas non-branded queries show organic growth because they reflect how new users find the site without any initial intent to visit it. If you look at both groups together, you see an average position and a click-through rate that matches neither group.
An example: if clicks rise in the fall, that may be due to a billboard campaign after which more people search for the company name. That has little to do with search engine optimization. Conversely, a good guide page can bring in new visitors while branded searches decline seasonally; in the total, it looks like standstill. Only the split shows which trend lies behind the total, and with it whether the work on content and technology reaches new customers.
A suitable metric is the share of non-branded clicks in all clicks. It tells you what portion of visitors from Google Search find the website through a topic rather than its name. The absolute numbers belong next to it, because a share can also rise when branded searches fall. The Search Console help recommends focusing more on trends in impressions and clicks than on position alone anyway.
How the metric is calculated
The share needs context because search behavior is shifting. Half (50 percent) of internet users aged 16 and over in Germany at least sometimes use an AI chat instead of a classic web search (Bitkom), while 47 percent still search exclusively the classic way (Bitkom). As for the AI summaries that search engines show above the results, 24 percent often use them without looking at the search results (Bitkom), while 43 percent click on the results to form their own opinion (Bitkom). For general questions in particular, which means non-branded queries, impressions can therefore grow while clicks stagnate. The article on AI Overviews and website visibility describes what role AI summaries play for visibility.
For your assessment, this means: if non-branded impressions rise while clicks stay flat, the website is visible for more topics but is clicked less often. Then it is worth looking at the title, the description and whether the page offers an answer that goes beyond a summary. The article measuring website success shows how search figures can be connected with inquiries and revenue.
Your own regex filter as a cross-check
Before the automatic filter, a regular expression was the usual way to strip out branded searches. Google brought regex filters for queries and pages to the performance reports on April 7, 2021, and explained them explicitly with the example of a company name and its abbreviations (Google Search Central Blog). The filter uses RE2 syntax, looks for partial matches anywhere in the query by default and is not case-sensitive; to change that, put (?-i) at the start of the expression.
The automatic filter does not make your own expression redundant. It serves as a cross-check: where the two numbers differ noticeably, the system classifies queries differently from your list, and those queries are worth a look. The regex filter also works in every property, including those without the automatic filter, and it can be combined with comparison mode. For several terms, separate the values with a vertical bar and put them in parentheses.
# Branded queries: company name, spelling variants, old name, own product
(companyname|company name|compnayname|old company name|productname)
# Cross-check: the same list with the option for non-matching
# shows all queries without these terms- In the Non-branded view, export the queries with the most clicks from the last three months.
- Collect company names, old names, abbreviations, owner names and your own product names, including the misspellings from the table.
- Join the terms with a vertical bar and put them in parentheses; spaces are part of the pattern.
- Apply the expression as a custom query filter and compare the total with the Branded view.
- Store the expression with a date in your own documentation and extend it when new products, locations or names come along.
| Feature | Branded / Non-branded filter | Custom regex filter |
|---|---|---|
| Basis | internal, AI-assisted Google system | your own list of terms in RE2 syntax |
| Misspellings and languages | recognized without anyone entering them | only what is in the expression |
| Own products without the company name | recognized if they clearly belong to the website | only what is in the expression |
| Availability | top-level properties with enough queries and impressions | every property |
| Maintenance | none, occasional misclassification possible | update for new names and products |
| Search Console API | not listed in the Search Analytics documentation | through the includingRegex and excludingRegex operators |
For analysis outside the interface, the Search Console API offers the includingRegex and excludingRegex operators in RE2 syntax; a request returns up to 25,000 rows (Google Search Console API). The API documentation does not list a branded query filter so far. If you want to continue the split in your own reports or spreadsheets, you therefore still need your own expression.
The year-over-year comparison at year end
At the end of the year, the obvious question is what the website achieved over the past year. When analyzing drops in search traffic, Google advises choosing the last 16 months in the date filter so that recurring effects such as holidays or seasonal trends become visible (Google Search Central). For the direct comparison, the same guide recommends comparing the last 3 months with the same period a year earlier (Google Search Central) and going through all tabs: queries, pages, countries, devices and search appearance.
When comparing two periods, the help recommends weekly or monthly granularity because it neutralizes the day-of-the-week effect that arises when, for example, a Monday is compared with a Saturday. The Branded and Non-branded views can be combined with the date comparison: set the filter first, then choose the comparison in the date filter. The table then shows a difference column that you can sort by to find the biggest winners and losers among pages and queries.
One question should be settled beforehand: do the branded filter values in your own property reach back into the previous year? The Search Console help states that the filter provides data from March 11, 2025 (Google Search Console) and calls that day its introduction. This does not match the announcement of November 20, 2025, which dates availability for all eligible sites to March 11, 2026 (Google Search Central Blog). Whether the automatic filter covers the previous year can therefore only be settled in your own property: a look at the last 16 months in the Branded view shows from when values are available. If they are missing for the comparison period, your own regex filter steps in, since it covers the entire stored period.
To keep the comparison readable later, events belong in the chart. You add custom annotations by right-clicking the chart, with a date and a note of up to 120 characters (Google Search Console). Each property can hold up to 200 annotations (Google Search Console), and they are deleted automatically after 500 days (Google Search Console), shortly after the end of the 16-month window. Annotations are hidden in comparison mode and in the 24-hour view, though; they appear in the plain time series, for example over the last 16 months. All users of the property can see annotations, so names, phone numbers or other personal information do not belong in them.
- Relaunch or move of the website with the date of the switch
- Renaming of the company, a location or a product
- Start and end of billboard, radio or ad campaigns that may trigger branded searches
- Publication of new service or guide pages
- Changes to the titles and descriptions of important pages
- Public holidays, company vacations and seasonal peaks that shift demand
A relaunch deserves special attention because it changes addresses and page titles; the article website relaunch without ranking loss explains what matters. For online shops, the Christmas season shifts the figures as well. If you advertise delivery dates in December, the article delivery by Christmas covers the legal questions; for the analysis, what counts is that campaigns and seasonality appear on both sides of the year-over-year comparison.
If you want to compare values beyond the 16 months, you have to save them. Google names the Search Analytics API and bulk data exports for this, which let you pull data and store it in your own systems (Google Search Central). For smaller websites, a monthly export of both views into a spreadsheet is often enough, supplemented by the non-branded share and a line with the month's annotations.
Turning the analysis into a routine
The filter delivers its value when it is read regularly and in the same way. The default view of the performance report shows clicks and impressions for the past three months (Google Search Console). For the monthly review, it is enough to set the period to the previous month and open both views one after the other. Once a quarter, add the comparison with the same period of the previous year, and once a year, look at the full 16 months.
- Check the property: does it show the filter with Branded and Non-branded?
- Open the Non-branded view and note the previous month's clicks and impressions.
- Open the Branded view, note the same values and calculate the non-branded share from them.
- In the Pages tab of the Non-branded view, check which pages bring in new visitors and which have lost them.
- Take a spot check for misclassifications and extend your own regex filter if needed.
- Record the month's important events as annotations in the chart.
Pages with many non-branded impressions but few clicks turn into concrete tasks. The Search Console help suggests updating the page's title and description or aligning the content better with the queries it appears for. Keyword research shows which terms matter; a content audit clarifies which older pages should be revised or merged.
For local businesses, the split has a special feature: queries for a service with a location are non-branded queries and therefore the field where local SEO works. If they rise after you have maintained your Google Business Profile, that is a sign of impact; it only becomes reliable when checked against calls and inquiries. If, on the other hand, you are mainly found by name, there is still room to grow with new customers, even if the total number of clicks looks good.
If you hire a provider for search engine optimization, you can make the non-branded share a shared metric and place it next to the absolute numbers in every report. The article checking SEO or GEO offers lists the questions to settle before hiring. For the coming year, the split provides the baseline values for setting goals in your online marketing plan for 2027.
Access for service providers
Related Articles
Delivery by Christmas: What the Promise Means in Your Shop
Delivery by Christmas: when a shop's delivery line becomes a promise, what Section 323 BGB allows after delays and how to set cut-offs per shipping method.
Unit Prices in Online Shops: Meeting Section 4 PAngV
Unit prices in online shops: when Section 4 PAngV applies, which unit Section 5 sets, which exemptions exist and how variants, feeds and product data follow.
Online Marketing 2027: Plan Channels, Budget, Calendar
Plan online marketing for 2027: a channel mix by maturity, a budget with base load and reserve, an annual calendar with seasonal peaks and metrics per channel.