Investigate page competition

Use page-query data to find page switching and possible intent overlap without overclaiming cannibalization.

Directional evidence Automatable now Advanced

Page competition occurs when more than one URL from a site earns impressions for the same query or query family. SEO teams often call this “keyword cannibalization,” but multiple-page visibility is not automatically harmful and Search Console does not report a cannibalization status.

The correct question is: Are these pages usefully serving different needs, or is unstable/overlapping targeting preventing the site from presenting a clear best page?

What Search Console can actually show

For a visible query:

Open Performance → Search results.
Select a complete date range and Search type.
Open Queries.
Click the query to filter the report to that exact value.
Open Pages.

The Pages table shows URLs that received reported data for the query during the selected period.

It does not prove that:

Every listed URL appeared simultaneously in one result set.
The URLs competed with one another on every impression.
Google was confused.
Consolidating them will improve rankings.
The visible query rows are exhaustive.

Results vary by time, location, device, history, and result layout. A 90-day table can combine different pages shown on different days and to different audiences.

When multiple pages are normal or beneficial

Keep distinct pages when they satisfy distinct user needs, such as:

A category page and a specific product page.
A definition and an advanced implementation guide.
Documentation and a commercial solution page.
A news article and an evergreen explainer.
Regional or language variants served to appropriate audiences.
Brand results with useful sitelinks.
Separate image, video, or rich-result experiences.
Several genuinely relevant pages for a broad query.

Google can show several results or sitelinks from one domain when useful. Shared query visibility is evidence to review, not an error to eliminate.

Signals that deserve investigation

A stronger page-competition case usually combines several signals:

Two or more pages serve essentially the same intent.
The preferred page changes repeatedly for the same stable query and segment.
Clicks and impressions fragment across near-duplicate pages.
A weaker or outdated page replaces a strategically preferred page.
Neither page performs as well as the previous single destination.
Titles, headings, content, and internal anchors communicate nearly identical purposes.
Canonical signals conflict or Google selects an unexpected canonical.
The wrong locale, product, or funnel-stage page appears.
Site architecture gives similar prominence and anchor text to several overlapping pages.

One shared query, by itself, is weak evidence.

Start with a query, then expand to a query family

An exact query is easy to inspect but can be noisy. Users express the same need through many variants. Build a reviewed family such as:

project management software
software for project management
project planning tool
project management platform

Use regex for a small transparent family or export query-page data for clustering. Preserve meaningful differences such as free, enterprise, template, login, location, and model number; they can signal a different intent.

Google’s language matching systems can relate a page to many query variants without the exact phrase appearing repeatedly. Do not create or merge pages solely based on literal wording.

Build a query-page matrix

For repeatable analysis, export or query these dimensions:

date + query + page + device + country

Keep dimensions only as granular as the available volume supports. Excessive dimensions can produce sparse top-row results.

A useful matrix contains:

Field	Purpose
Query family	Stable analytical unit
Page	Competing destination
Clicks and impressions	Materiality
CTR	Result interaction clue
Average position	Supporting visibility evidence
First/last date seen	Persistence
Weekly leading page	Detects switching
Device/country	Separates audience-specific behavior
Google-selected canonical	Detects duplication/canonical issues
Intended page	Business/content decision, not a GSC field

The Search Analytics API can request Page and Query together, but it returns top rows and does not guarantee every row. BigQuery bulk export is preferable for large, durable analysis.

Measure page switching

Aggregate the matrix by week rather than treating daily movement as meaningful.

For each query family and segment:

Identify the page with the most impressions each week.
Count how often the leading page changes.
Measure each page’s share of impressions and clicks.
Compare total family performance before and after switches.
Require a minimum impression baseline.

Example directional measures:

page impression share = page impressions / family impressions
page click share = page clicks / family clicks

Do not sum page impressions and call the result property impressions without considering aggregation. Page-grouped data can count several URLs separately while the property-level chart counts the property once.

Switching is more concerning when the same country, device, and query family are stable and total clicks decline. It is less concerning when different pages lead in different countries, devices, or sub-intents.

Classify the situation before acting

Complementary visibility

The pages have distinct purposes and both earn relevant clicks. Keep them. Improve clarity, internal navigation, and titles so users understand the difference.

Intent mismatch

A page appears for a query it only partly satisfies while a better page exists. Clarify content focus, title, headings, and internal anchor context. Ensure the preferred page is crawlable, indexable, internally linked, and complete.

Near-duplicate consolidation problem

The pages contain substantially the same primary content or represent URL variants. Review redirects, rel="canonical", sitemaps, internal links, and URL parameters. Canonicalization is the relevant technical concept here.

Editorial overlap

Two indexable pages independently target the same need but contain different fragments of useful information. Decide whether they can sustain distinct intents. If not, combine the best material into one stronger page and redirect the retired URL.

Expected page transition

A new page is replacing an old page after a migration or consolidation. Temporary overlap can be normal while Google recrawls and processes signals. Verify redirects/canonicals and monitor rather than repeatedly changing direction.

Audience segmentation

Different localized or device URLs appear for different users. Validate hreflang, canonical, redirect, and mobile configuration before treating it as competition.

Check canonicalization separately

Google clusters pages when it finds the same or very similar primary content and chooses a representative canonical. Most Search Console Performance data is credited to that canonical.

For each suspected duplicate:

Inspect both URLs with URL Inspection.
Record the user-declared and Google-selected canonical.
Check status codes and redirects.
Check rel="canonical" in rendered HTML or HTTP headers.
Check sitemap inclusion.
Check whether internal links point consistently to the preferred URL.

Canonicalization and editorial competition are not the same problem. Do not add a canonical from one genuinely distinct page to another merely to force a preferred ranking. Canonical is for duplicate or substantially similar alternatives.

Evaluate content intent

Compare the pages side by side:

Primary audience and task.
Funnel stage.
Main answer or product.
Title and H1.
Unique information or experience.
Query families earning impressions.
Internal links pointing to the page.
Conversion or engagement goal.
Freshness requirements.

If you cannot describe a durable user-centered difference in one sentence, the pages may overlap. If the difference is clear and useful, shared keywords are not enough reason to merge them.

Check internal linking and site structure

Google uses links to discover pages and as a relevance signal. Every important page should receive crawlable internal links from relevant pages, with concise descriptive anchor text.

For the preferred page:

Link from the relevant hub and supporting articles.
Use natural anchors that describe its distinct purpose.
Avoid forcing the same exact keyword into every anchor.
Remove or update links to retired URLs.
Keep breadcrumbs, navigation, and sitemaps consistent.

For complementary pages, use internal links to explain the relationship: overview to detail, category to product, beginner to advanced, or problem to solution.

Choose an intervention

Keep both and differentiate

Use when both pages satisfy durable distinct intents.

Clarify titles, headings, and introductions.
Remove accidental duplicated sections.
Strengthen unique evidence and examples.
Cross-link with descriptive context.
Monitor query families separately.

Consolidate and redirect

Use when one page can fully satisfy the shared intent and preserve the useful content of both.

Choose the strongest long-term URL.
Merge unique useful material.
Publish and validate the destination.
Permanently redirect the retired URL.
Update internal links, canonicals, and sitemaps.
Monitor Page Indexing and Performance.

Correct duplicate signals

Use consistent redirects, canonical hints, internal links, HTTPS, and sitemap URLs for duplicate variants. Google may still select a different canonical because these signals are hints rather than commands.

Improve the intended page

Use when the correct page exists but is incomplete, poorly linked, or misaligned with the query’s user need. Improve it for people rather than stuffing the query phrase.

Retire or noindex

Use only when a page should genuinely not appear in Search. noindex removes a page from eligible results; it is not a precision ranking tool for choosing among valuable pages.

Actions to avoid

Merging every pair of pages that shares a query.
Creating one page for every keyword variation.
Using rel="canonical" between pages with distinct primary content.
Deleting a page without preserving valuable unique material or links.
Judging competition from a one-day or low-impression sample.
Assuming the URL with the best average position is always the best business destination.
Treating a manual search from one device/location as historical proof.
Repeating exact-match internal anchors unnaturally.

Validate after a change

Record a baseline before implementation:

Total query-family clicks and impressions.
Each page’s share.
Leading page by week.
CTR and average position.
Index and canonical state.

Add a Search Console annotation, then wait for recrawling and enough complete data. Google says changes can take weeks or longer to settle.

Success is not merely “only one URL remains.” Evaluate:

Total family clicks and impressions.
Whether the intended page is now stable.
Whether distinct sub-intents were lost.
Post-click outcomes.
Canonical and indexing correctness.

What this investigation can conclude

It can establish:

Which reported pages received data for a visible query or family.
Whether the leading page changes over time.
How clicks and impressions are distributed.
Whether the pattern is device/country-specific.
Whether canonical selection is unexpected.

It cannot prove:

That pages appeared together on every result page.
That one URL directly suppressed another.
Why Google selected a result.
That consolidation will improve rankings.
Every query affected, because privacy and top-row limits apply.

Use “possible harmful overlap” until the evidence and content review justify a stronger conclusion.

Investigation checklist

The query has meaningful volume.
A query family, not only one phrase, is reviewed.
Date, device, and country are controlled.
Page switching is sustained across weeks.
Total family performance is measured.
Complementary intent is considered.
Canonicals and redirects are inspected separately.
Internal links and architecture are reviewed.
The intervention preserves unique user value.
Results are measured after recrawl with enough data.