Short answer: Keyword cannibalization is when two or more of your own pages compete for the same search query, trade places in the results, split the clicks, and leave neither page ranking as high as one focused page might.
It is not automatic, though. Two pages ranking for one query can be a healthy result that takes up more of the page, so the job is to separate genuine overlap from harmless overlap before changing anything. Often the right answer is to do nothing.
What is keyword cannibalization?
Search engines pick the page they think best answers a query. When you have two pages that answer it about equally well, the engine may alternate between them, show one at some times and the other at others, or rank both lower than you would expect from either alone. The pages are, in effect, competing with each other instead of with the rest of the results.
The key word is competing. Sharing a query is not the same as competing for it. A category page and a review both appearing for best running shoes can be the best case, because together they occupy more of the results. The problem begins when they trade places, split clicks, and neither one breaks through.
Two of your pages ranking for one query is not automatically a problem. Treat overlap as a prompt to look, not as a defect, and expect that some of what you find should simply be left alone.
How do you tell real cannibalization from harmless overlap?
Look at the competing pages' positions and click-through together, and then ask what each page is for. The data shows overlap. It cannot show intent, so when the addresses do not make the purpose obvious, ask the person who owns the content rather than guessing.
| Signal | Probably real cannibalization | Probably fine |
|---|---|---|
| Position | Both pages sit in a similar range, for example 5 to 15, and neither breaks through | One page clearly dominates and the other picks up a small tail |
| Intent | Both serve the same intent, such as two posts answering one question | They serve different intents for an ambiguous query, such as a product page and a support article |
| Click-through | Poor on both relative to their position | Healthy on the page that matters for that query |
| Purpose | No reason for both to exist | One deliberately targets a different audience or funnel stage |
Mind the volume too. On a large site, the long tail produces overlaps that are technically real and practically meaningless. Raise the impression threshold until the list is small enough to review, and start at the top, where the most traffic is at stake.
When is overlap fine to leave alone?
More often than the cannibalization articles suggest. If one page clearly wins and the other collects a small tail, merging them risks the tail for no gain. If the two pages serve different needs behind an ambiguous query, they are each doing their job. If one page exists for a non-search reason, such as a campaign landing page, its overlap is a cost you chose.
Doing nothing is a legitimate outcome, and a good report says so plainly. A review that finds three real conflicts and eleven harmless overlaps is more useful than one that flags fourteen and recommends eleven unnecessary rewrites.
Duplicate content is a different question
Two near-identical pages is a duplicate-content situation, which Google's guidance handles with canonical URLs and consolidation. Cannibalization is broader: the pages may be quite different but still answer the same query. The tools overlap, but the diagnosis is not the same.
Google: Consolidate duplicate URLsWhat should you do about real cannibalization?
There are four options, in rough order of preference. Pick the lightest one that solves the problem.
Consolidate
Merge the weaker page into the stronger one and permanently redirect the old address. Best when both pages genuinely serve one intent.
Differentiate
Rewrite one page to target a clearly different query. Best when both pages should exist but currently overlap.
De-target deliberately
Remove the competing terms from the weaker page and keep the page. Best when it has a purpose other than search.
Do nothing
A legitimate result, and often the correct one.
Which page to keep matters. A sensible default is the page that already ranks better, because rank is the harder half to earn back. A page at position 4 with fewer clicks is usually a better target than one at position 28 with more. Override that when the lower page is more useful, more current or converts better, or when the higher page ranks on the strength of the domain despite thin content. Backlinks also matter, and Search Console does not show them.
Never delete without a redirect
Dropping a URL that has backlinks, or that other pages link to, causes a worse problem than the one you were fixing. A permanent redirect to the page you are keeping preserves what the old URL earned.
Google: Redirects and Google SearchHow do you find the overlaps on your own site?
In Search Console, list the queries where more than one of your pages received impressions, group them by query, and sort by total impressions. It is possible by hand, but tedious on a big site, which is why we packaged it as an agent skill. It can be installed into Claude Code and other coding agents, and it applies the judgement above rather than declaring every overlap a fault. The guide has the exact steps.
Read the cannibalization skill guideThe skill source is public on GitHub.
View the skill source on GitHubThe overview lists every available skill and how to install them.
Install agent skillsIf your traffic has fallen rather than just overlapped, start with the drop diagnosis instead.
Why did my organic traffic drop?Common questions
What is keyword cannibalization?
It is when two or more pages on your own site rank for the same query and compete with each other, splitting clicks and impressions so that neither reaches the position a single focused page might. It is a real issue only when the pages serve the same intent and trade places without either breaking through.
Is having two pages rank for the same keyword always bad?
No. A category page and a review both appearing for one query can occupy more of the results. It becomes a problem when they swap places, split clicks, and both sit outside the top results. When one page clearly dominates, or the pages serve different intents, leave it alone.
How do I fix keyword cannibalization?
Choose the lightest fix that works. Consolidate by merging the weaker page into the stronger one with a permanent redirect, differentiate by rewriting one page for a different query, or deliberately remove the competing terms from the weaker page. Doing nothing is also a valid outcome.
Should I delete one of the competing pages?
Not without a redirect. A deleted URL loses whatever backlinks and internal links it earned, which is worse than the overlap. If you consolidate, permanently redirect the old address to the page you keep, and preserve any useful content from the old one.
Sources
Related guides
Want it done for you
Want this done and re-checked every month? Our monthly maintenance re-runs your buying questions in ChatGPT, Perplexity and Google, ships the fixes that keep you out of the answers, and reports what moved. Or run the free check on your site first at advancelabs.dev/tools/audit.
See how it works