How to Fix Keyword Cannibalization

Keyword cannibalization is something you confirm with data before you change anything: two or more of your own URLs collecting impressions for the same query in Search Console, with Google showing first one and then the other. Once confirmed, the fix depends on how the pages relate. Merge pages that answer the same need, separate pages that serve different needs, and fix architecture when the wrong type of page keeps appearing. Set expectations early: removing self-competition gives your best page a fair fight. It doesn’t win the fight for you.

Confirm it with the Performance report

In Search Console’s Performance report, use the filters: click “+ Add filter”, choose Queries, pick “Queries containing” and enter the query. Then switch to the Pages tab. Cannibalization looks like more than one of your URLs drawing impressions for that query, and, over a longer date range, positions that trade places as Google alternates between the pages. The trading pattern is a practitioner’s sign, not a Google definition, and a useful one. That filter also catches longer variants; to match one exact query, use a regular expression anchored at both ends, in the form ^your query$.

Three cautions about reading that view:

• Filtered totals are partial. Google’s filtering help says filtering by query or URL can affect the report totals because of data truncation and the omission of anonymized queries. Compare the pages’ shares of the filtered data, not the filtered total against your site total.
• One URL with impressions isn’t cannibalization. If only one of your pages draws meaningful impressions for the query, you have an ordinary ranking problem.
• Shared words aren’t shared queries. Two pages that use the same term but rank for different queries aren’t competing.

On a large content library, check query families instead of single queries. The filtering help describes a regular expression search that matches several queries with variable parts, which lets you group, for example, every “how to fix” variant of a topic and see which pages divide it.

Let the results page assign intent

When two pages serve arguably different needs, don’t settle it in a meeting. Search the head term and see which kind of page leads: product grids, or comparison and explainer content. Then search the longer variant and do the same. The type of page Google shows for each query suggests which of your pages should own it. If the head term shows product grids and your editorial guide is trying to rank there, move the guide to the comparison query where editorial content leads, and point the commercial page at the head term.

Classify the relationship

The fix follows from the relationship between the pages. Getting this step wrong is expensive: merging a different-intent pair destroys a page that should exist, and separating true duplicates leaves two thin pages where one strong page belonged.

Same need: merge

Two pages answer the same query in the same way. Keep the stronger one, move anything useful from the other into it, and redirect the retired URL with a permanent redirect.

Different needs: separate

Two pages serve different goals that happen to share a term, such as a buying guide and a product page. Keep both and make each one’s focus unmistakable.

Wrong page type: fix the architecture

Google keeps showing the category page when a product page should answer, or the reverse. The pages can be fine while the signals around them point the wrong way.

Separate by removing overlap

Once each page has its query, make the signals stop contradicting it. The title, the H1 and the page’s focus should each point one way. Remove commercial words such as “best” and “top” from the page you decided should be the editorial resource, and remove editorial framing from the page meant to rank for the commercial query. If both titles still say “best”, they still pitch the same need.

The work is subtraction as much as addition: take out what made the pages interchangeable.

Architecture and internal links

When the wrong page type keeps appearing, look at how your site links to each page. Links help Google judge which pages are relevant, according to its link best practices. Give the page that should own the query links from related content, with anchor text that describes it, and check whether your navigation links point at the competing page instead.

Two structural fixes to check:

• Product pages target product queries. A single product page trying to rank for the broad category term is misaligned by design.
• Filtered and parameter variants shouldn’t compete. Point their canonical at the clean URL, or keep them out of crawling if they don’t need to be indexed.

Set honest expectations

Changes register as Google recrawls the pages and can settle on one page per query, so judge the result over weeks of data, not days. Watch the same filtered Performance view you used to confirm the problem, and look for positions to stop trading places.

On the mechanics of a merge: Google’s guide to consolidating duplicate URLs ranks redirects first among the signals that influence which URL becomes canonical, with rel="canonical" next. When one page is retired for good, the redirect is the clearer instruction.

Sequence matters. Separate first, merge last. Merging retires a page, so keep it for true duplicates; when the pages serve different needs, separating them can keep both earning.

Frequently asked questions

How do I know it’s cannibalization and not just a low ranking?

Filter the Performance report to the query and open the Pages tab. Cannibalization shows several of your URLs drawing impressions for that query, with positions trading places over time. One URL ranking poorly is an ordinary ranking issue.

Should I just delete one of the pages?

Not as the first move. Merge only true duplicates that serve the same need. When the pages serve different needs, separate them so both can keep working.

Will fixing cannibalization make me rank higher?

It removes self-competition and can steady which page Google shows. The remaining page still has to win on its own merits against other sites.