Organic search can look healthy while answering the wrong commercial question. A service business may receive more clicks from queries Google associates with its brand, including names, variants, misspellings, or distinctive related products and services. That traffic is useful, but it should be analyzed separately from queries Google classifies as non-branded. The split describes Google's classification of a query, not what a person already knew or how the person first encountered the business.
Google announced a branded queries filter on November 20, 2025 and said on March 11, 2026 that it was available to all eligible sites. The current help page separately describes history beginning March 11, 2025; preserve that documentation discrepancy. The classifier uses names, variations, misspellings, and associated products or services, with eligibility and accuracy limits.
The useful question is whether both buckets change for understandable reasons and whether non-branded discovery reaches the right services, proof, and inquiry paths. This guide builds a repeatable baseline without pretending Search Console is a complete attribution system.
What branded and non-branded actually mean
Google defines branded queries as searches that include a brand name, variations or misspellings of it, and brand-related products or services. Non-branded queries are the remaining searches. In the Performance report, the filter can be applied across web, image, video, and news search types. The selected view limits familiar Search Console metrics such as clicks, impressions, average position, and click-through rate to the chosen query class.
This is not the same as new versus returning customer. A first-time prospect may hear the name from a neighbor and perform a branded search. A long-time customer may search non-branded for an adjacent service. A branded click can still be commercially important, and a non-branded impression may never become a visit. The classification describes the query, not the person's relationship to the business.
| View | Useful for | Does not prove | Common next drill-down |
|---|---|---|---|
| Branded | Brand-demand trend, navigation demand, reputation checks | Customer loyalty or campaign causation | Homepage, contact, review, and location pages |
| Non-branded | Discovery trend, service and problem demand, content reach | A new customer or qualified lead | Service, location, guide, and proof pages |
| Combined | Overall Search presence and technical anomaly checks | Growth beyond existing awareness | Split the data before explaining change |
Why the new filter is better—and where it stops
Before the filter, analysts commonly approximated brand traffic with query strings or regular expressions. That method can miss spelling variants, brand-associated products, abbreviations, and languages. Google's classifier reduces that maintenance burden and gives eligible sites a consistent interface for comparing query types. The related Insights card also summarizes the click split between branded and non-branded traffic.
Google also documents important constraints. The filter may be unavailable to sites with a low number of impressions and is not available for sub-properties. Some queries may be classified incorrectly. Search Console omits anonymized queries from query tables for privacy and stores only the most important rows, so displayed query lists are not exhaustive. Filtering by query can change the relationship between chart totals and table rows.
If the native filter is unavailable, label the property ineligible. A versioned regular-expression fallback can preserve brand terms, variants, exclusions, and an unclassified bucket, but it is an analyst rule—not Google's classifier. Do not compare it directly with a later native series without a bridge-period audit.
Classification-quality check
- Sample exposed queries from both buckets each quarter, including the highest-click rows and a random lower-volume set.
- Label obvious false positives, false negatives, ambiguous product names, locations, and generic words that overlap the brand.
- Report reviewed-row counts plus clicks and impressions represented; calculate any observed error rate only within that weighted reviewed sample.
- Define the ambiguous bucket in advance when an analyst regex fallback is used, and preserve it instead of forcing every term into brand or non-brand.
- Keep the original Google bucket in raw data even when an analyst adds a corrected review label.
- Record that the Search Analytics API does not currently document the native branded filter; use a governed manual export unless the current API contract explicitly adds it.
Build a baseline that survives ordinary volatility
Begin with a long enough period to see the business's normal pattern. For a seasonal service, a year-over-year comparison is usually more interpretable than comparing a busy month with the immediately preceding slow month. Weekly or monthly aggregation can reduce distraction from weekends and day-level noise. Keep date ranges aligned and note holidays, severe weather, promotions, migrations, outages, major content releases, and changes to paid or offline campaigns that may influence branded demand.
| Field | Required record | Why it matters |
|---|---|---|
| Population | Exact property identifier, property type and scope, total portfolio, plus a frozen logical priority-page cohort | Prevents property or URL changes from masquerading as performance change |
| Window | Exact current and comparison dates with equal duration and completeness | Controls seasonality and partial-period bias |
| Filters | Search type, country, device, native brand-filter availability, page rule, and aggregation mode | Makes the export reproducible |
| Metrics and ratios | Clicks, impressions, CTR with impressions as denominator, classified-click mix, visible-query ratio, and page-population metadata | Keeps each calculation tied to its actual population |
| Coverage | Anonymized-query limitation, export row limit, preliminary dates, and earliest native-filter history | Prevents visible rows from being called the complete query set |
| Annotations | Release, outage, campaign, weather, and methodology events | Separates observed timing from causal proof |
Baseline capture
- Record the property type and confirm it represents the full production site rather than a narrow sub-property.
- Preserve the exact property identifier and whether the native branded filter was available in that property and period.
- Use comparable date ranges and preserve the exact start, end, and comparison dates.
- Export branded and non-branded views separately before adding other filters.
- Capture clicks, impressions, CTR, and average position, but lead the narrative with trend and page relevance rather than position alone.
- Repeat the split by page, device, country, and relevant Search type.
- Annotate releases and external events without claiming timing proves causation.
- Record when the filter is unavailable or when low volume makes a segment unstable.
- Define materiality as both an absolute and relative change threshold, with a documented commercial override for high-value pages.
- Keep raw exports, logical-page register, and reporting definition so a later analyst can reproduce the comparison across URL changes.
Search Console offers custom annotations and weekly or monthly views. Because annotations are shared with property users, keep them free of personal information and use a compact date, change-class, route-group, and release-ID convention linked to the underlying record.
Move from a sitewide split to service-line evidence
A sitewide non-branded increase can be commercially weak if it lands on an irrelevant article or a service the business does not prioritize. Report both the total portfolio and a frozen logical-page cohort. Map old and new URLs to the same logical page where redirects or canonical migrations preserve purpose; list genuinely new, removed, split, or merged pages separately. Selecting only current winners creates survivor bias and can hide a priority page that disappeared.
Search Console generally assigns performance to the canonical URL, not necessarily the visited URL. Use URL Inspection and the site's canonical rules before interpreting a zero or assigning credit to a template.
Service-line review sequence
- Choose one commercially meaningful service family rather than grouping the entire site into a single SEO result.
- Freeze the priority-page population before comparison and list additions, removals, and canonical changes separately.
- Filter to non-branded queries and review the landing pages receiving impressions and clicks.
- Inspect the individual queries Google exposes, while acknowledging anonymized and truncated rows.
- Classify visible queries by service need, location, urgency, comparison, troubleshooting, or informational intent.
- Check whether the ranking page is the intended page and whether it contains current proof and a usable inquiry path.
- Compare the same service family across a like period before deciding that a change is material.
- Send landing-page sessions into the separate lead-evidence workflow instead of calling the Search Console click a lead.
Diagnose the shape of change, not just the total
A useful report explains which component moved. If branded impressions rise while CTR falls, the business may be appearing for a broader set of brand-related searches, a result presentation may have changed, or the page title and result context may not match expectations. If non-branded impressions rise without clicks, the business may be entering consideration but appearing too low or with weak result messaging. Average position is an aggregate and should not be treated as a precise rank tracker.
Device and geography can expose operational issues. Mobile non-branded clicks may grow while the mobile call path is broken. The country dimension describes where the search originated; it is not a city, radius, or service-area report. A location page receiving country-level impressions does not prove demand inside or outside the operating territory. Image or video visibility may increase without sending users to the intended conversion page. Each pattern is a prompt for inspection, not a verdict.
| Observed pattern | Reasonable next question | Overclaim to avoid |
|---|---|---|
| Branded clicks up | Did brand demand, navigation need, or a campaign change? | SEO created all new demand |
| Non-branded impressions up | Which pages and query groups entered more results? | Qualified leads increased |
| Non-branded CTR down | Did position, device mix, result format, or title relevance change? | Google penalized the site |
| Clicks up, CRM flat | Are calls, forms, qualification, and source context captured? | Search traffic has no value |
| One page down | Did the canonical, intent, content, availability, or SERP context change? | The whole domain lost authority |
Report decisions, limitations, and commercial next steps
An executive report does not need every query. It needs the branded and non-branded trend, the service families responsible for material movement, the fixed and changed page populations, data limitations, and the next decision. Decompose the movement into impressions, CTR, page mix, device/country mix, and canonical or population changes before telling a content story. A good next decision might be repairing a canonical, improving a result title, strengthening a service page, consolidating duplicate intent, or repairing an inquiry path.
Do not turn the split into an agency scoreboard. Public relations, paid media, referrals, reviews, offline reputation, competition, location, accessibility, content, and result features can all shape the buckets. Use the report to expose dependencies and choose a bounded action.
| Output | Current | Comparison | Required qualifier |
|---|---|---|---|
| Branded and non-branded metrics | Clicks, impressions, CTR | Same metrics and filters | Native filter or versioned analyst fallback |
| Portfolio and logical-page cohort | Totals plus additions/removals | Same logical pages | Canonical and URL mapping preserved |
| Query visibility | Visible rows, clicks and impressions represented | Same fields | Anonymization and truncation stated |
| Lead coverage | Accepted leads with usable source / all accepted leads | Same denominator | Aggregate comparison, not user matching |
| Observed layer | Example | Permitted conclusion |
|---|---|---|
| Search Console | Non-branded clicks to a frozen set of six emergency-service pages rose from 240 to 300 year over year; impressions rose 30% while CTR declined | Discovery increased for this fixed page set; result mix or position still needs inspection |
| Analytics | Aggregate landing sessions to those pages rose from 210 to 252; consent and blocking create a coverage gap | Recorded sessions moved in the same direction, not that every click became a session |
| CRM | Accepted inquiries rose from 18 to 20; usable source context fell from 8/18 (44%) to 6/20 (30%), leaving 14 current records unknown | A small accepted-lead increase was observed while source coverage worsened; attribution remains incomplete |
| Decision | Mobile call testing found one broken sticky-call action | Repair and retest the customer path before attributing the trend to content |
The downstream connection is an aggregate reconciliation, not a person-level join. Search Console does not expose a user identifier that can be matched to an analytics session or CRM record. Align comparable landing-page populations and periods, show coverage gaps at each layer, and leave unattributable leads unattributed.
Use a monthly operating cadence
For most service businesses, a monthly review with a deeper quarterly service-line analysis is more useful than daily reaction. Check for technical anomalies immediately, but let ordinary search demand accumulate before rewriting strategy. Use the same property, filter definitions, date logic, and service families each cycle. Record any methodology change so apparent growth is not merely a changed report.
The branded filter should make reporting more honest. It separates queries Google classifies as branded from those it classifies as non-branded, while preserving the reality that both groups can matter and that the classifier can be wrong. It also makes a useful commercial conversation possible: Is the business maintaining brand-associated demand, expanding discovery for priority services, sending that discovery to the right pages, and capturing enough downstream evidence to decide what to improve next?
Answer with trends and documented uncertainty. Search Console does not show every query, identify a person, partition Maps activity, or close the loop to revenue. Within those boundaries, the split keeps organic reporting from hiding behind one blended number.
Source ledger
These sources support the operating guidance above. Platform behavior and documentation can change, so volatile implementation details should be rechecked before a rollout.
- Introducing the branded queries filter in Search Console — Google Search Central. Published November 20, 2025; updated March 11, 2026 for general eligible-site availability.
- Performance report: Dimensions and data groupings — Google Search Console Help. Documents classification, eligibility, history, canonical aggregation, anonymized queries, and truncation.
- Performance report: Common tasks and use cases — Google Search Console Help. Official workflows for brand awareness, page analysis, and evaluating performance changes.
- Performance report: About the data — Google Search Console Help. Defines aggregation, preliminary data, and export behavior.
- Using Search Console and Google Analytics data for SEO — Google Search Central. Official boundary between Search Console search activity and on-site Analytics behavior.
- Adding context to your Search Console data with custom annotations — Google Search Central. Published November 17, 2025; introduces custom performance-chart annotations.
- Introducing weekly and monthly views in Search Console — Google Search Central. Published December 10, 2025; introduces longer time aggregation and notes export changes.
- Search Analytics: query — Google for Developers. Current API dimensions and filtering contract; the native branded-query classifier is not documented as an API dimension.
- Understand your Business Profile performance — Google Business Profile Help. Official definitions and coverage boundaries for calls, directions, website interactions, and Ads-influenced performance.