Keyword cannibalization: how to find it and how to fix it

What is keyword cannibalization?

Keyword cannibalization is when two or more pages on the same website compete for the same search query, so Google keeps choosing between them and neither ranks as well as one clear page would. You see it as impressions for one query split across several of your URLs, or as the ranking URL changing from week to week while the position stays stuck.

It usually happens by accident. A blog post and a service page both target “roof repair cost”. A 2023 guide and a 2025 update both stay live. A category page and a product page share a title. Each page alone might be fine; together they send mixed signals about which one you want to rank.

The cost is mostly indirect. Links, internal links and engagement are spread across two pages instead of concentrated on one. Google has to pick, and it may pick the weaker page. And you spend effort improving one page while Google shows the other.

When keyword cannibalization is not a problem

Two of your pages appearing for one query is only cannibalization when both get a real share of the impressions and neither is clearly the right answer. Plenty of overlap is harmless, and “fixing” it can do damage. Leave these alone:

• Brand queries. When people search your name, two or three of your pages on the results page is more space for you, not competition.
• One page already dominates. If one URL holds nearly all the impressions and another appears now and then, there is nothing to consolidate.
• Different intents. A “what is X” guide and an “X pricing” page can both show for a broad query. If each wins its own specific searches, they are complementary.
• Long-tail overlap. Any two related pages share some rare queries. Judge by the queries that matter to you, not by how many overlap.
• Fragment rows. Search Console can report /page/#section separately from /page/. Those are one page.
• Search operators. Queries like site:example.com match every page by design.

How to find keyword cannibalization in Search Console

To find keyword cannibalization, open a query in Search Console's Performance report and switch to the Pages tab: every URL listed there appeared for that query, with its own impressions, clicks and position. Two URLs with meaningful impressions inside the top 20 positions is the signal to look closer.

1. Pick the queries that matter

   In Performance → Search results, set the last 3 months, sort the Queries table by impressions, and work down your important non-brand queries. Cannibalization on a query nobody searches is not worth fixing.

2. Open the query, then the Pages tab

   Click the query to filter the report to it, then open the Pages tab. Note each URL's impressions and position. Google's Performance report help explains why the numbers look odd: the Queries table is aggregated by property, the Pages table by page. Two of your pages in one result count as one impression on the Queries tab and two on the Pages tab.

3. Work out each page's share

   Divide each URL's impressions by the total. One page with 90% and another with 10% is not a problem. Two pages at 60/40 or 50/50, both on page one or two, are competing.

4. Check whether the ranking URL swaps

   Compare the Pages tab across two or three consecutive months, or add the date dimension through the Search Analytics API. A query where one month's top URL is the guide and the next month's is the service page is the clearest sign Google cannot decide.

5. Confirm the page you want is healthy

   Before changing anything, run the page you plan to keep through URL Inspection. It should be indexed, and Google's selected canonical should be the page itself.

For a whole-site sweep, the Search Console API can return the query and page dimensions together, so you get every query-page pair in one export and can sort for queries with more than one URL. Our keyword audit guide walks through that export step by step.

How to fix keyword cannibalization: a decision table

Fixing keyword cannibalization starts with choosing one main page for the query; the technical fix then depends on whether the other page needs to stay live and whether it serves a different intent. Choose the page with the stronger links, better content and the URL you want long term, not just the one ranking today.

What Google says about redirects and canonicals

Google's documentation calls a redirect a strong signal that the target should become canonical, and rel="canonical" a strong signal that the named URL should become canonical; a sitemap entry is only a weak signal. Google's canonicalization guide adds three points that matter for cannibalization:

• Use redirects to retire duplicates, canonicals to keep them live. Google recommends a redirect when you are getting rid of a duplicate page, and rel="canonical" when the duplicate URLs need to stay accessible.
• Do not use noindex to choose a canonical. Google advises against it within one site, because noindex removes the page from Search entirely instead of passing its signals to the main page.
• Do not use robots.txt for it either. A URL blocked in robots.txt can still be indexed without its content, and Google cannot see a canonical on a page it is not allowed to crawl.

On redirect type, Google's redirects guide says permanent redirects (301 or 308) make Google show the target in results and act as a canonical signal, while temporary ones (302, 303, 307) keep the old URL in results. It recommends server-side redirects whenever possible and treats JavaScript redirects as a last resort. Use a permanent redirect only when you are sure it will not be reverted. You can confirm a redirect returns a single 301 hop with our redirect checker.

For internal links, Google's link best practices say Google uses links both to find pages and to understand how relevant they are, and that anchor text should be descriptive and natural, without cramming in keywords. That is why pointing your related pages at the main page, with the query as anchor text, helps Google settle on it.

What to do after the fix

After a merge, redirect or re-target, update your internal links and sitemap, then give Search Console a few weeks before judging the result.

• Update internal links to point at the main page directly, not at the redirected URL. A link through a redirect still works, but it is an extra hop for crawlers and visitors.
• Remove the redirected URL from your XML sitemap. List only the final, canonical URLs.
• Request indexing for the main page in URL Inspection if you changed it substantially.
• Record the date. Compare the query's Pages tab four to eight weeks later: the goal is one URL holding most of the impressions, at a better average position.

To stop it happening again, keep one sheet that assigns one page to each important query before you write anything new. Our guide to a master keyword list shows how.

How MrKeyword's keyword cannibalization checker works

MrKeyword checks your connected Search Console data for queries where two or more of your pages genuinely compete, and filters out the false alarms described above. A query counts as cannibalization only when at least two pages each have 5 or more impressions and 10% or more of the query's impressions at an average position of 20 or better. It skips search operators like site:, brand queries (matched against your domain name), and overlaps where one page already holds 80% or more of the impressions. #section rows are folded into their page.

The same detector feeds several places in the app:

• This Week, the SEO opportunities inbox, lists “Your pages compete for …” items with the number of pages, the shared impressions and the best position, plus a suggested page to keep.
• Keyword Strategy groups queries into topics. When competing queries hold at least 20% of a topic's impressions, its suggested next step is to merge the competing pages into one main page.
• Content Ideas has a Keyword Cannibalization card showing each competing page's position, impressions and share, with a button to add a task.
• Rank Tracking can send a Cannibalization alert, in the app or by email, for queries where several URLs got impressions in the last 28 days. Each tracked keyword can also have a target page; when Google ranks a different URL, the keyword shows a Wrong page badge, which is often the first sign of cannibalization.
• The MCP server's daily issues list includes keyword cannibalization, so an AI assistant reading your account sees it too.

To add internal links toward the page you keep, the internal link suggestions find sentences on your other pages that already contain the target's own Search Console queries. See pricing for what each plan includes.

What MrKeyword does not do

• It does not merge pages, add redirects or edit canonicals for you. It shows which pages compete and suggests one to keep; you make the change on your site.
• It only sees what Search Console reports. Queries Google anonymizes for privacy are not in any keyword row, so cannibalization on very rare queries cannot be detected.
• It does not judge intent for you. Two pages flagged as competing may serve different needs; read both before merging.
• It does not check other people's sites. Cannibalization detection needs your own connected Search Console property; there is no public cannibalization checker for any domain.
• It does not scrape Google results. Positions are Search Console averages over a period, not a live rank at one moment.

A related problem on the other end of the scale is a page with no internal links at all. See how to find and fix orphan pages.

Questions and answers