Keyword cannibalization: diagnosing it in Search Console and fixing it

Keyword cannibalization happens when several pages on the same site target the same search need. Search engines then have to choose which one to show, and they often change their mind. The result is a set of pages that each rank worse than a single, focused page would, with impressions and links split between them. It is common on sites that have published content for years without a plan: three posts about the same topic written by different people, a category page and a blog post fighting over a commercial query, or near-identical location pages.

It is also one of the most over-diagnosed problems in SEO. Two pages appearing for the same query is not cannibalization by itself. This guide shows how to tell the difference with data, and how to fix the real cases.

What real cannibalization looks like

Three signs together point to a genuine problem:

1. The same query is split across two or more of your URLs in Search Console, with each getting a meaningful share of impressions.
2. The ranking URL changes over time: one week page A, the next page B, then A again, often with the position dropping each time it switches.
3. The pages serve the same intent. A reader satisfied by page A would be equally satisfied by page B.

If only the first sign is present, it may be healthy. A guide and a category page both appearing for a broad query, each in the top five, is a win, not a problem. If the pages serve different intents, the fix is to make the difference clearer, not to merge them.

Finding candidates in Search Console

The Performance report can show which pages appear for a query. For a single query, filter by the query (exact match), then switch to the Pages tab. For finding candidates across the site, the interface is slow, and the API or bulk export is better: pull data with both query and page dimensions, then group by query and count distinct pages.

A simple shortlist rule:

• Queries with at least a meaningful number of impressions (pick a threshold that fits your site, such as 100 per month).
• Two or more of your pages each receiving at least 20% of that query's impressions.
• Best average position worse than about 5.

Then look at those queries over time. In the interface, filter by the query, set the date range to several months, and compare pages week by week. A pattern of alternating pages with jagged positions confirms the switching. The average position guide explains why the averages need care: a page that appears rarely at a good position can look better than it is.

Common causes

Cause	Example
Repeated blog topics	"How to choose a standing desk" (2021), "Standing desk buying guide" (2023), "Best standing desks" (2025)
Category and article overlap	A category page and a "best X" article both targeting "X"
Tag and category archives	Tag "standing desks" and category "standing desks" listing the same posts
Location or variant pages	Near-identical service pages for neighbouring towns
Same anchor text to different pages	Internal links saying "standing desks" point to three different URLs
Product variants on separate URLs	Each colour as a separate page with the same description

Choosing the fix

Merge and redirect

When pages serve the same intent, merging is usually the strongest fix. Choose the page with the best combination of links, traffic and URL, move the best content from the others into it, rewrite it into one coherent piece, then 301-redirect the others to it. Update internal links to point directly at the merged page.

The merged page should be better than any of the originals, not just longer. Remove repetition, keep the most specific details and examples, and check that it still matches what the results page rewards.

Differentiate

When pages could serve different intents but currently overlap, sharpen each one. Turn one into the beginner guide and the other into the advanced comparison; or make the article clearly informational ("how to choose") and the category clearly transactional. Change titles, headings and introductions so the difference is obvious, and adjust internal anchors accordingly.

Re-point internal links

Sometimes the content is fine and only the signals are mixed. If internal links use the same anchor text for several pages, search engines get no clear answer about which page is the main one for that phrase. Pick one target per topic and make internal links consistent. This alone can settle a switching pattern. The internal linking guide covers anchor strategy.

Noindex or remove

For thin archive pages (tags duplicating categories) or old posts with no unique value, noindex or removal is appropriate. The pruning guide covers how to decide between noindex, delete and redirect.

Canonical tags, rarely

A canonical from one page to another tells search engines they are duplicates. That is correct for true duplicates such as parameter variants, but between two genuinely different articles it is a weak and often ignored signal. Prefer merging or differentiating.

A worked example

A hypothetical home-office furniture shop finds that the query "standing desk for small spaces" splits impressions between a 2022 blog post, a 2024 buying guide and a category filter page. Positions swing between 7 and 18. All three target the same need: someone looking for a compact standing desk to buy.

Decision: the category filter page ("compact standing desks") becomes the target for the commercial query, with a short buying section added under the grid. The 2024 buying guide is rewritten to focus on measuring and choosing for small rooms, linking to the category with the anchor "compact standing desks". The 2022 post, which overlaps the guide almost entirely, is merged into the guide and redirected. Internal links across the blog are updated so "compact standing desk" always points to the category.

Measured over the next two months against the previous two, the category settles into a stable position and the guide begins ranking for the "how to choose" queries it now targets. Clicks are compared for the combined set of URLs, not just one page, so the effect of the redirect is visible.

What not to do

• Do not delete pages without redirects. Removing the weaker page throws away whatever links and history it had. Redirect it to the page that absorbs its content.
• Do not merge pages that serve different intents just because they share a keyword. A "what is" explainer and a "best X" buying guide may both mention the same phrase and still deserve to exist separately.
• Do not stuff the chosen page with every variant of the keyword after merging. Write it for the reader; search engines understand synonyms.
• Do not judge after a week. Rankings move around while redirects are processed. Give it a month.
• Do not chase every split. Long-tail queries with a few impressions spread across pages are noise. Focus on queries that matter to the business.

Preventing it

• Keep a keyword-to-page map, as described in the keyword research guide, and check it before commissioning new content.
• Search your own site (site:example.com topic) before writing about a topic; update an existing page rather than writing a new one when possible.
• Use consistent internal anchor text for each target page.
• Review queries with multiple ranking URLs every quarter.

Measuring the result

After a fix, compare equal periods before and after for the query group, summing clicks across all the affected URLs. Expect a short dip while redirects are processed, then a stable position for the chosen page. If the chosen page still swaps with another URL after a month, look for remaining signals pointing elsewhere: internal links, sitemap entries, or another page you missed. Tracking positions day by day, as DeskSEO does when Search Console is connected, makes the switching pattern and its end easy to see.