If your organic traffic is growing, how much of it is people already searching for your business by name? And how much is new demand your content earned? Until now, answering that in Search Console meant writing your own regex (a pattern-matching rule). The new search console branded queries filter does the split for you. And custom annotations let you mark the site changes that explain the bumps in your charts.
Google launched both features within a few days of each other. Custom annotations arrived in the Performance report on 17 November 2025. The branded queries filter was announced on 20 November 2025 and is rolling out slowly over the following weeks. Here is what each one does, who gets them, and how I would build them into your reports.
What Google Announced in November 2025
These are two separate features. But they solve the same problem: Performance report charts that show what happened without explaining why.
- Branded queries filter (20 November 2025): splits your Search performance into branded and non-branded queries using an AI-assisted classifier (a system that sorts queries into groups). It comes with a new card in the Insights report that shows how your clicks divide between the two.
- Custom annotations (17 November 2025): lets you add short notes to dates on the Performance report chart. So you and your team can see what changed and when.
Neither feature adds new data. They organise the data you already have. That makes it easier to read, and easier to explain to someone who does not live in Search Console.
How the Search Console Branded Queries Filter Works
The filter sorts every query that brought your site impressions into one of two groups. You can then see clicks, impressions, CTR and average position for branded queries only, or for everything else.
What Google Counts as a Branded Query
Google’s announcement on the Search Central blog says a branded query is one that includes your brand name. But the sorting goes further than a simple text match. It is built to catch:
- Your brand name in any language.
- Common typos and misspellings of the brand.
- Searches for products or services unique to your brand, even when the brand name is not in the query. Google’s own example is “Gmail” being treated as a branded query for Google.
That last point is the big difference from a homemade regex. A regex only finds the words you thought of. The classifier can pick up a signature product name you might never have added to your pattern.
It Is AI-assisted, so Check It
Google says the sorting is done by an internal AI-assisted system and that it can sometimes get things wrong. At launch, there is no way to tell it a query is in the wrong group or to add your own brand terms.
I treat the split as a very good starting point, not a source of truth. Before you put it in a client report, export a sample of each group and scan for clear mistakes. Watch out for:
- Brand names that are also common words, for example a bakery called “Daily Bread”.
- Product names that overlap with general category terms.
- Searches for competitor brands. These are not your branded traffic, even though they contain a brand.
Where You Will Find It
The filter lives in the Performance report for Search results. Google has said it works across the search types in that report: web, image, video and news. You can combine it with your other filters, such as country, device or a single page.
The Insights report gets a matching card. It shows the share of your total clicks that come from branded and non-branded queries. That card is the quickest way to see how much your site depends on people who already know you. It is the first thing I would look at once the feature appears.
Who Gets the Branded Filter, and When
This is the part most people will run into first, so let me be precise. Google has set three conditions:
- Top-level properties only. The filter is on top-level properties. A URL-prefix property for a subfolder, or a separate subdomain property, will not show it.
- Enough query volume. Sites need enough search volume for the split to mean something. A small or very new site may not see the filter at all.
- Gradual rollout. Google is releasing it over the coming weeks. If your property qualifies but you can’t see it yet, it may just not have reached you.
If you only have URL-prefix properties, this is a good reason to verify a Domain property. It covers every protocol and subdomain, which also gives you a fuller picture in every other report. I set up a Domain property on every site I work on for this reason.
The Fallback: A Regex Brand Filter
If the filter has not reached your property, the old method still works. In the Performance report, add a Query filter, choose Custom (regex), and list your brand variations with a pipe between them. Search Console uses RE2 syntax. Adding (?i) at the start makes the match ignore upper and lower case. I share more patterns in my Search Console regex examples.
(?i)acme|acmeshop|acme shop|akme|acme-toolsSet the filter to “Matches regex” for branded traffic and “Doesn’t match regex” for non-branded. It is less clever than the AI classifier, but it is transparent: you know exactly what is in each group. Some teams will keep a regex version running next to the new filter for that reason. Honestly, I plan to do the same for a while.
Using the Branded Filter for Reporting
The real value is not the filter itself. It is the questions it lets you answer quickly and the same way every time.
1. Report Non-Branded Growth as Your SEO Headline
Branded clicks mostly reflect demand that already exists: advertising, word of mouth, returning customers, offline activity. Non-branded clicks are where content, technical work and links show up. If you report one blended number, a good month for brand awareness can hide a bad month for SEO, and the other way round.
A simple monthly structure works well:
- Non-branded clicks and impressions, month on month and year on year.
- Branded clicks, reported separately as a sign of brand demand.
- The branded share of total clicks, from the Insights card, tracked over time.
2. Read Branded CTR and Position Differently
Branded queries usually have much higher CTR and a top average position, because people searching your name want your site. Mixing them into site averages makes your overall CTR look healthier than your discovery traffic really is.
Once you split them, a falling branded CTR is a useful warning sign. It can mean a competitor, a review site or a marketplace listing is taking clicks on your own name. Non-branded CTR is where title and snippet work tends to show results. It is also where rewriting titles and meta descriptions for pages that rank but do not get clicked tends to pay off.
3. Find Content Opportunities in Non-Branded Queries
With brand traffic removed, the Queries tab becomes a cleaner list of topics people link with your site. Sort non-branded queries by impressions. Look for terms with high impressions, low clicks and an average position between about 5 and 20. Those are pages that are close to earning traffic, and they are usually where I start.
Switch to the Pages tab with the non-branded filter on. If two URLs keep showing up for the same non-branded queries, you may have a keyword cannibalization problem worth fixing.
4. Separate Brand Campaigns From SEO Results
If your business runs TV, radio, sponsorship or social campaigns, branded search should rise while they run. The filter lets you show that lift without it inflating your SEO numbers. It also helps paid search teams see how much of the brand demand organic results already catch.
5. Check for Drops in the Right Bucket
When traffic falls, the first question should be: which group fell? A drop in non-branded traffic points to rankings, indexing or a search update. A drop in branded traffic points to falling brand demand, seasonality or something outside SEO. In my experience, that one split can save hours of digging.
How Custom Annotations Work
Search Console already shows some system annotations on charts, for example when Google flags a data issue. Custom annotations add your own notes next to them.
The key details from Google’s help page:
- Where: on the Performance report chart. Annotations stay on their date when you apply filters. But they do not show in comparison mode or the 24-hour view.
- Length: up to 120 characters per note.
- Limit: up to 200 annotations per property.
- Retention: annotations older than 500 days are deleted automatically.
- Editing: you cannot edit an annotation. To change one, delete it and add a new one.
- Who sees them: everyone with access to the property. Owners and full users can add and delete. Restricted users can only view.
Because annotations are shared, Google advises treating them as business data. Keep personal information, such as names, phone numbers or addresses, out of them. Write “Checkout redesign live”, not a colleague’s name.
The full details are in Google’s help page on custom annotations in Search Console.
What to Annotate (and What to Skip)
With a 200-note limit and 500-day retention, annotations work best as a log of events that could really move search performance. Good candidates:
- Site changes: migrations, redesigns, URL structure changes, CMS or hosting moves, large batches of redirects.
- SEO work: title and meta rewrites, internal linking projects, structured data added, large content updates or removals.
- Content launches: a new section, a large batch of articles, a product range going live.
- Google changes: confirmed core or spam updates, and reporting changes. The mid-September drop in desktop impressions after Google dropped the num=100 parameter is exactly the kind of event worth marking. Then nobody mistakes it for lost rankings months later.
- Business events: sales, product recalls, PR coverage, brand campaigns, holidays and seasonal peaks.
- Incidents: outages, accidental noindex tags, robots.txt mistakes, and the date each was fixed.
Skip anything routine, like single blog posts or small copy edits. If everything is annotated, the chart becomes noise and you will hit the limit faster.
A Simple Naming Convention
120 characters is short, so agree a format with your team. Starting each note with a tag makes the chart easy to scan. This is the style I like:
[SITE] Migrated /shop/ to /store/, 301s live
[SEO] Rewrote titles on 40 category pages
[GOOGLE] Core update rollout started
[BIZ] Black Friday sale starts
[FIX] Removed accidental noindex on /blog/Keep a change log outside Search Console as well, for example with GA4 annotations. Annotations disappear after 500 days and cannot be edited. So Search Console should not be your only record for year-on-year comparisons.
Putting Both Features Together
The two features are most useful together. Annotations stay visible when you apply filters. So you can look at non-branded performance with your change log on the same chart.
Take a site that relaunched its category pages in October. With all queries combined, the chart may look flat, because an autumn dip in branded demand cancels out the gain. Apply the non-branded filter and the lift after the annotated launch date becomes visible. Apply the branded filter and you can show the dip was seasonal, not caused by the relaunch.
The same approach helps after any search update. When a new feature or model reaches Google Search, such as Gemini 3 arriving in AI Mode, an annotation on that date lets you watch whether non-branded impressions or CTR shift over the next few weeks.
Limits to Keep in Mind
- Not every property qualifies. Subfolder and subdomain properties, and low-volume sites, will not get the branded filter.
- You can’t edit the sorting. You cannot fix mistakes or add brand terms, so check the output before you rely on it.
- Anonymised queries still apply. Search Console leaves out some rare queries for privacy. So query-level totals do not always add up to the chart total. That is true for both groups.
- Annotations are not permanent. They are capped at 200 per property and removed after 500 days.
- Annotations do not appear in comparison mode. If you compare date ranges, keep your change log handy separately.
Conclusion
The search console branded queries filter finally makes the most important SEO reporting split a built-in view: brand demand on one side, discovery traffic on the other. Custom annotations add the missing context, so a spike or dip on the chart comes with a reason attached. I think these are two of the most practical reporting features Google has added in a long time.
Your next step: check whether you have a Domain property, add annotations for the main site changes of the last few months, and rebuild your monthly report around non-branded growth. If you are still finding your way around the tool, start with my guide to the Search Console reports that matter.
