Branded vs non-branded performance

Separate existing brand demand from discovery and non-brand organic growth.

Directional evidence Partially automatable Intermediate

Separating branded from non-branded queries prevents existing brand demand from being mistaken for broader organic discovery. Branded performance reflects searches that Google classifies as including your brand, domain, or a brand-specific product or service. Non-branded performance covers more general searches.

The split is useful, but it is not a perfect audience or attribution model. A branded query does not prove the searcher is an existing customer, and a non-branded query does not prove the searcher is new.

What the two segments mean

Branded queries

Typical branded queries contain:

The company or site name.
The domain name.
A distinctive product or service owned by the brand.
A common misspelling or variation.
Brand plus another need, such as acme pricing or acme returns.

These searches often indicate prior awareness, navigation, support needs, reputation research, or demand created by marketing outside organic Search.

Non-branded queries

Typical non-branded queries express a category, problem, question, comparison, or local need without a known brand identifier. They are a useful indicator of discoverability beyond explicit brand-name searches.

Non-brand growth can show that pages are earning visibility for broader topics. It does not measure unique new users or incremental customers; connect Search Console to GA4 and business data for landing-page outcomes.

Google’s branded-query filter

Google provides Branded and Non-branded options inside the Performance report’s Query filter. Google says classification is AI-powered and informational only. It does not affect rankings, and some queries can be mislabeled.

Current documented boundaries are important:

Data begins on March 11, 2025, when the filter was introduced.
The filter can expose up to the normal 16-month Performance history, but no branded classification exists before that introduction date.
It works across Web, Image, Video, and News Search types.
It is unavailable for sites with a low number of impressions.
It is unavailable for sub-properties such as a path-only https://example.com/blog/ property.
Rollout-dependent related Insights features might not appear for every otherwise eligible property at the same time.

Do not promise the control in a client workflow until you have confirmed it on the exact property.

Use the built-in filter

Open Performance → Search results.
Select the Search type and a complete date range on or after March 11, 2025.
Add a Query filter.
Choose Branded or Non-branded.
Review Clicks and Impressions first.
Inspect Pages, Queries, Countries, and Devices within the filtered segment.
Use a date comparison to evaluate change.

The filter applies to the full report view, so page metrics represent clicks and impressions attributed to the selected query class.

Validate Google’s classification

AI classification should be quality-checked, not accepted as ground truth.

Sample both sides

Review at least:

The highest-click branded rows.
The highest-impression branded rows.
A sample from the long tail available in the interface or export.
The highest-click non-branded rows.
Ambiguous product and category terms.
Common misspellings and translated/transliterated variants.

Record false positives and false negatives. Recheck when the brand portfolio changes, a product becomes genericized, or Google updates the feature.

Common classification ambiguity

A short brand name that is also an ordinary word.
A product name shared with a category.
Distributor or partner brands.
Former company names after a rebrand.
Founder names.
Local-script and transliterated spellings.
Queries containing a competitor and your brand.
Product model identifiers that do not contain the company name.

There is no universal correct answer for every reporting purpose. Decide whether the goal is Google’s classification, navigational demand, demand for owned products, or a strictly rule-based finance metric.

Build a transparent regex alternative

Use a custom query regex when the built-in filter is unavailable, wrong often enough to impair decisions, or insufficiently transparent for a recurring KPI.

A basic brand family might be:

\b(acme|acme corp|acmecorp)\b

Use Matches regex for brand and Doesn’t match regex for a rule-based non-brand view.

Build the brand dictionary

Consider separate, reviewed lists for:

Company and site names.
Domain with and without the top-level domain.
Unique product names.
Abbreviations.
Previous names.
Common misspellings.
Localized names and transliterations.

Add a term only when it uniquely indicates the brand for the reporting objective. A broad term such as apple, notion, or square can create many false positives without boundaries or context.

Use boundaries and anchors

Search Console regex is a partial, case-insensitive RE2 match by default. Word boundaries can reduce substring errors:

\bacme\b

An entire-query navigational set can use anchors:

^(acme|acme corp|acme\.com)$

This narrower pattern excludes acme pricing. It answers “pure navigational queries,” not “all branded queries.” Name the segment accordingly.

Version the definition

Store:

The exact expression.
Plain-language inclusion rules.
Known ambiguous terms.
Positive and negative examples.
Owner and review date.
The date a new term was added.

Changing the regex changes the metric. For a historical trend, either apply the new definition to all compared dates or disclose the break.

Built-in classification versus regex

Factor	Google branded filter	Custom regex
Setup	Simple	Requires a maintained dictionary
Misspellings and variants	AI may infer them	Must be explicitly represented
Transparency	Classification logic is not exposed	Rule is inspectable
Availability	Eligibility and property limits	Available wherever query regex works
Historical start	March 11, 2025	Within available Performance history
Consistency	Can change as classification evolves	Stable until the rule changes
Ambiguous names	Can be mislabeled	Can also be wrong if rules are broad
Automation	Not equivalent to a standard API dimension	Can be implemented as query filters downstream

Do not silently mix these two definitions in one trend line. Label them Google-classified brand and rule-based brand.

Hidden queries prevent a perfect reconciliation

Search Console removes some queries for privacy and truncates lower-importance rows. Google includes anonymized queries in unfiltered chart totals, but applying a query filter removes them from the filtered dataset.

Therefore:

branded clicks + non-branded clicks may not equal unfiltered clicks

The same problem applies to a regex brand/non-brand pair. Do not create an “unknown brand” number by subtracting filtered values unless you clearly explain that it includes privacy and reporting effects, not just unclassified demand.

Metrics to report

For each segment, track:

Clicks.
Impressions.
CTR.
Average position as supporting context.
Share of reported clicks.
Absolute click change.
Year-over-year and recent-period change.
Top landing pages.
Top visible query families.
Country and device concentration.

Calculate share using values with the same filtering boundary:

reported non-brand click share = non-brand clicks / (brand clicks + non-brand clicks)

This is often more internally consistent than dividing by the unfiltered total, but it still describes reported classified clicks rather than all organic clicks.

Interpret common patterns

Branded clicks rise, non-brand clicks are flat

Possible explanations include stronger brand demand from advertising, PR, offline activity, existing customer behavior, or improved branded result performance. It is not evidence that non-brand SEO expanded.

Non-brand impressions rise, clicks are flat

The site may be appearing for more queries or at lower positions. Segment by page and query, then inspect CTR and position. This can be an early discovery opportunity rather than a failure.

Non-brand clicks rise with stable branded demand

This is compatible with broader organic discovery. Confirm that gains are distributed across meaningful query families and pages, not caused by a formerly branded term being reclassified.

Brand rises after a campaign

Treat campaign timing as context. Search Console can show the query and click change but cannot attribute brand lift to one channel. Use campaign, geographic, and time-based evidence from other systems.

Brand falls while total clicks stay flat

Non-brand discovery may have offset weaker brand demand. That can be healthy for visibility but still signal a brand or retention problem. SEO and brand teams may reach different valid conclusions from the same split.

Analyze landing pages within each segment

The most useful question is often not “What is our brand share?” but “Which pages serve each type of demand?”

For branded queries, inspect:

Homepage and login pages.
Pricing and product pages.
Support, returns, and contact content.
Review and reputation pages.
Unexpected pages winning navigational queries.

For non-branded queries, inspect:

Informational and comparison pages.
Product and category discovery pages.
Localized landing pages.
Pages with high impressions but weak CTR.
Pages whose growth is concentrated in one query or country.

If the homepage receives many non-brand impressions, inspect its query relevance. If a support article captures important brand-commercial queries, evaluate whether the intended product page better satisfies the need.

What the split cannot tell you

Neither method can determine by itself:

Whether a searcher already knew the brand.
Whether the user is new or returning.
Whether a non-brand click is incremental.
Which marketing channel caused a brand search.
Search volume outside your site’s impressions.
Competitor share of voice.
Revenue, leads, or lifetime value.

Combine GA4, CRM, campaign, and competitive data for these questions. Keep the Search Console segment as a description of your reported Google Search performance.

Reporting checklist

The classification method is named.
The March 11, 2025 built-in-filter boundary is respected.
Property eligibility is confirmed.
Visible queries from both segments are sampled.
Ambiguous terms and known errors are documented.
Query privacy effects are disclosed.
Absolute changes accompany shares and percentages.
Search type, dates, country, and device are consistent.
The rule is versioned when regex is used.
“Non-brand” is not presented as “new customers.”