The Performance report: complete guide

Master metrics, dimensions, filters, date ranges, search types, and report aggregation.

Automatable now Beginner

The Performance report explains how a Search Console property appeared and earned clicks in Google Search. It is the primary source for reported Google Search clicks, impressions, click-through rate, and average position.

It answers questions about your property’s visibility, not the whole search market. It does not expose competitor impressions, exact search volume, conversions, crawl activity, or a fixed rank for every keyword.

Open and scope the report

Open Performance → Search results for the selected property. Before interpreting any number, record:

Property.
Search type.
Date range and comparison period.
Active filters.
Selected metrics.
Table dimension.
Whether the newest data is preliminary.

The default view shows recent Web Search data. A number copied without this scope cannot be reproduced reliably.

Metrics

Clicks

A click generally counts when a user selects a supported Google Search result link that takes them outside Google to the property. Actions that keep the user inside Search are not always counted as external clicks. Result-type-specific rules can differ.

Use clicks as the primary traffic outcome in this report. They are not identical to GA4 sessions: redirects, canonical attribution, consent, tracking, bots, time zones, and metric definitions create legitimate differences.

Impressions

An impression counts when a link or content item is eligible to be considered seen under the rules for its result type. Some items must be scrolled into view; others can count when their containing result page loads. Compound elements and carousels have specific counting rules.

Impressions describe reported visibility for your property, not total query search volume. If your property never appeared, Search Console has no impression to report.

CTR

Click-through rate is:

CTR = clicks / impressions

CTR varies naturally by position, query intent, brand familiarity, device, country, and result layout. A sitewide average is not a universal benchmark. Diagnose CTR within comparable query/page/device/country groups.

Average position

Position is the topmost position occupied by the property or selected page in a result, averaged across impressions. Lower numbers are closer to the top.

Average position is not:

A fixed daily keyword rank.
The position of every URL from the site.
Directly comparable across very different query mixes.
Proof of why clicks changed.

When multiple results from the property appear, property aggregation normally uses the topmost one. In compound search features, position can refer to the containing element or the item according to that result type’s methodology.

Dimensions

Dimensions group the table. Filters restrict the population shown in the chart and table.

Queries

Queries are the search terms associated with your property’s reported impressions and clicks. Use them to identify demand themes, brand versus non-brand performance, intent, and page relevance.

Some queries are omitted for privacy, and Search Console stores and displays only important top rows. Visible query rows therefore do not necessarily sum to chart totals.

Pages

Pages group data by the final URL linked from the result after redirects, with most data assigned to Google’s selected canonical URL. Use URL Inspection when the reported page is unexpected.

Selecting Pages changes the table to page aggregation; the chart remains property-aggregated.

Countries

Country groups data by the country where the search originated. It is not a language, nationality, or residency dimension.

Devices

Device groups searches as Mobile, Desktop, or Tablet. It represents the searcher’s device category, not whether the site has a separate mobile page.

Search appearance

Search appearance groups supported result types and features for which the property recorded impressions, such as particular rich or product results. Available values depend on the site’s actual data.

Search-appearance table data is page-aggregated. A URL can have more than one appearance in the same result set, and filtered clicks are attributed to a URL rather than uniquely to a (URL + appearance) interaction. Treat appearance click analysis with those rules in mind.

Dates

Dates show time trends. The interface supports Hourly granularity for the 24-hour view and Daily, Weekly, or Monthly granularity for appropriate longer ranges. Weekly and Monthly views help expose sustained trends without daily noise.

The newest points can be preliminary and are drawn with a dotted line. Use finalized complete days for formal comparisons.

Search types are separate datasets

Use the Search type control to choose:

Web.
Image.
Video.
News.

Search type is not the same as Search appearance. Image Search data and an image-related appearance inside another context are different reporting concepts. Search Console does not combine Search, Discover, and Google News performance into one browser table; their reports must be analyzed separately or exported.

Always keep Search type constant when comparing periods unless the comparison itself is explicitly about source mix.

Filters

Filters answer narrower questions. You can filter across several dimensions at the same time; those conditions use AND logic.

Example:

Country = Spain
AND Device = Mobile
AND Page contains /guides/

After applying those filters, choose Queries to see the visible queries for that exact population or Pages to see its landing pages.

Query and Page filters support literal containing/not-containing conditions and custom RE2 regular expressions. Query filters are case-insensitive. Regex matching is partial and case-insensitive by default unless anchors or (?-i) change it.

Applying any query filter removes anonymized queries from filtered totals. A filtered branded/non-branded split may therefore not add back to the unfiltered chart.

Comparisons

Comparisons place two values from one dimension side by side:

One date range versus another.
Mobile versus Desktop.
One country versus another.
One page versus another.
Two search appearances.

The table displays the values and a Difference column. Only one comparison can be active at a time, though ordinary filters can remain.

Use equal-length complete date ranges with the same weekdays. Match seasonal periods when demand varies. Weekly or Monthly granularity can make long comparisons easier to interpret.

Rare values on one side can appear as ~, meaning unavailable, not necessarily zero.

The chart and table do not always use the same aggregation

This is one of the report’s most important rules.

Property aggregation

The chart and table dimensions such as Query, Country, Device, and Date are generally property-aggregated. If several results from the same property appear for one query:

They can count as one property impression.
Multiple clicks after returning to the result set can count as one property click.
Position is the topmost property result.

Page aggregation

When grouping by Page or Search appearance, table rows are page-aggregated. Each unique page can receive its own impression, click, and topmost page position.

Consequently, the Pages table can sum to more impressions or clicks than the property-level chart. This is not automatically an error.

Do not compare a page-summed CTR or position with the chart as though the calculations were identical.

Canonical URL attribution

Google normally credits clicks, impressions, and position for duplicate variants to the Google-selected canonical URL. A user may click or land on a different URL while Search Console reports the canonical.

This affects:

HTTP/HTTPS or www variants.
Tracking parameters.
Mobile/desktop URL pairs.
International near-duplicates.
Migrated URLs and redirect chains.
Syndicated or duplicated content across properties.

Canonical attribution can move all visible performance to a different property if that property contains the canonical URL. Use URL Inspection and canonical diagnostics before declaring a duplicate URL to have zero demand.

Data availability and limits

History

The Performance report generally exposes up to 16 months of data. Feature-specific dimensions may have shorter histories; for example, Google’s built-in branded-query classification begins March 11, 2025.

Search Console is not a permanent data warehouse. Export data continuously when longer history is required.

Privacy and top rows

Google omits some queries to protect privacy and truncates less-important rows. Chart totals can exceed the sum of visible queries. The Search Analytics API also returns top rows and does not guarantee every underlying row.

Freshness

Recent data can be preliminary. The API can request fresh or hourly data and returns metadata indicating incomplete dates or hours in supported requests.

Time boundaries

Be careful when joining Search Console with systems using a different time zone. The Search Analytics API documents standard dates in Pacific Time, while the 24-hour interface view uses local-time behavior. Normalize dates before row-level comparison.

Reporting changes and anomalies

Google can change result types, counting methods, or logging. Check Search Console annotations and Google’s data-anomalies page when a chart break aligns with a documented reporting event.

A reliable analysis sequence

Select the correct property and Search type.
Choose a complete date range and an equivalent comparison period.
Start with Clicks and Impressions to establish whether traffic or reported visibility changed.
Segment by Page to locate the affected site area.
Select affected pages, then inspect Queries to understand demand and relevance.
Check Device, Country, and Search appearance for concentration.
Use CTR and Average position as supporting evidence within comparable segments.
Check canonical changes, site releases, seasonality, and reporting anomalies.
Export when thresholds, URL classification, or more rows are needed.
Label conclusions as observed fact, hypothesis, or confirmed cause.

Common questions and the correct view

Question	Configuration
Which pages earned the most Web clicks?	Search type Web → Pages → sort Clicks
Which queries show a selected page?	Click/filter the page → Queries
Which pages lost the most clicks?	Compare equivalent dates → Pages → sort click Difference
Is mobile CTR weaker than desktop?	Device comparison → control Page/Country/Query as needed
Which countries are growing?	Compare dates → Countries → sort click Difference
Which high-impression queries have low CTR?	Queries → export/filter with minimum impressions
How did `/guides/` perform year over year?	Page contains `/guides/` → compare dates
Did non-brand discovery grow?	Built-in non-brand or documented regex → compare dates
Which rich-result type performs best?	Search appearance, with aggregation/click caveats

Misleading interpretations to avoid

“Average position 8 means we rank eighth”

It is an impression-weighted average of the topmost result for the selected grouping across different searches and contexts.

“The query table is all searches”

Privacy filtering and top-row limits make it incomplete.

“Page rows should equal the chart”

They can use different aggregation.

“Higher impressions always mean improvement”

They can come from new low-position queries or different demand.

“Low CTR means a bad title”

Position, intent, device, brand, and result layout can explain the gap.

“A chart change proves our release caused it”

Search Console shows timing and association, not controlled causality.

Export, API, and BigQuery choices

Interface export

Use the Export control for the current chart and visible table configuration. Record active filters with the file because the rows alone may not explain their scope. Exported ~ and - values can appear as zero, so preserve availability meaning separately when important.

Search Analytics API

Use the API for repeatable filtered requests, multiple dimensions, and scheduled top-row extraction. The request supports dimensions, filter groups, aggregation type, search type, data state, and row paging. A single request can ask for up to 25,000 rows, but internal limits mean the API still does not promise all data.

BigQuery bulk export

Use bulk export for large sites that need the most complete available daily performance dataset, durable history, custom classifications, and joins with business data. It adds warehouse cost and governance responsibilities and does not remove privacy protections.

What the report cannot answer alone

The Performance report cannot directly tell you:

Which competitor replaced your result.
Total keyword search volume or market share.
Whether the page was crawled on a specific date.
Why Google chose a ranking.
What the user did after landing.
Revenue, leads, or retention.
Whether a title or content change caused the metric change.
Every query that produced an impression.

Use URL Inspection, Page Indexing, Crawl Stats, GA4, business systems, server logs, and competitive research as appropriate.

Important limitations

Treat Clicks and Impressions as the primary observed metrics; interpret CTR and Average position in context.
Preserve property, Search type, dates, filters, and aggregation with every export.
Do not reconstruct unfiltered totals from filtered query segments.
Do not infer language from country or new users from non-brand queries.
Do not treat a comparison as causal evidence.
Keep long-term exports because interface history is finite.

Used carefully, the report is a precise record of how Google reported your property’s Search visibility and clicks. Used without its boundaries, it can produce confident but false explanations.