Keyword cannibalization is when two or more pages on your own site compete for the same search query, so neither one gets the ranking a single stronger page would have earned.
It is also one of the most over-diagnosed problems in SEO. Two of your pages showing up for a query is often completely fine. This covers how to tell the harmless version from the damaging one, how to find real cases using nothing but Search Console, and when to leave it alone entirely.
What it actually looks like
The textbook description is two pages targeting one keyword. In practice you rarely set out to do that, and cannibalization arrives by accident.
- Topic drift. You write a broad guide, then a narrower one, and over a year the narrow one grows until it covers most of the broad one.
- Series posts. Four annual updates of the same subject, all still published, all still indexed.
- Category and tag archives ranking for the same term as the article they list.
- A product page and a guide for the same thing, where the query could reasonably want either.
Notice that none of these look wrong when you write them. They look wrong only in aggregate, which is why cannibalization surfaces during a content audit rather than during publishing.
The signal that matters is flipping, not sharing
This is where most diagnoses go wrong. Two of your pages appearing for one query is not, by itself, a problem. Google regularly shows more than one result from a site, and a guide and a product page can legitimately both be relevant.
The damaging pattern is flipping: Google keeps changing its mind about which of your pages to show for the query. One week it serves page A, the next week page B, and neither accumulates the stability that comes with a settled ranking.
You can see this in Search Console. Filter to the query, break down by page, and look across weeks rather than at a total. Steady split between two pages is usually fine. Alternating, where one page’s impressions collapse as the other’s rise and then they swap back, is the real thing.
The second damaging pattern is the wrong page winning. Your thin listing page ranks and your detailed guide does not, so the click lands somewhere that cannot satisfy the searcher.
How to find it with Search Console alone
You do not need a paid tool for this. The method below is free and reproducible.
- Open Performance and set the range to the last 12 months. Search Console keeps 16 months, so this is comfortably inside what it holds.
- Go to the Queries tab and pick a query that matters to you and earns real impressions.
- Click it to filter, then switch to the Pages tab. You now see every page of yours that earns impressions for that one query. This pairing is the whole technique.
- If two or more pages appear, check the trend. Change the date comparison to month by month and watch whether the split is stable or alternating.
- Repeat for your top twenty queries. Twenty is enough. Cannibalization on a query nobody searches is not worth an afternoon.
If you want this in bulk rather than query by query, export the Performance data with both query and page dimensions, then group by query and count distinct pages. The Search Console API returns up to 25,000 rows per request against a default of 1,000, so raise the limit or you will only see the head of your data.
Score it before you fix it
Not every case deserves work. Rate each one you find.
Fix now. The query earns meaningful impressions, two pages alternate across months, and neither reaches the top ten. This is costing you a ranking you could have had.
Fix if convenient. Two pages share the query steadily, one is clearly the better answer, and the weaker one is the one Google prefers. The cost here is a worse click, not a lost ranking.
Leave it. Both pages rank, or the split is stable and both are useful, or the query is tiny. Two of your results on page one is a good outcome, not a problem to solve.
Not actually cannibalization. The pages rank for different queries that merely look similar, or one of them is not indexed at all, in which case nothing is competing and the real issue is covered in crawled, currently not indexed.
The fixes, strongest first
Merge and redirect. Fold the weaker page into the stronger one and 301 the old URL. This is the right answer most of the time, because it consolidates both the content and whatever links the weaker page had earned. It is also the only fix that makes the problem permanently go away.
Differentiate the intent. If both pages deserve to exist, make them answer visibly different questions. Change the titles, the opening paragraphs and the headings so a reader and a crawler can both tell them apart in two seconds. Half-hearted differentiation leaves you with the same problem and more words.
Fix the internal linking. Often Google picks the wrong page because your own site points at it more. Count the internal links to each, then link deliberately to the page you want to win, using the query as the anchor where it reads naturally.
Canonicalise. Useful when both pages must stay reachable for users but only one should rank, such as a printable version or a near-duplicate variant. It is a weaker signal than a redirect and Google may ignore it.
Noindex the loser. The blunt option. It works, and it removes a page from search entirely, so keep it for archives and tag pages rather than for real content.
Whatever you choose, change one thing and then wait. These fixes take weeks to settle, and stacking three at once leaves you unable to say which worked.
When cannibalization is not your problem
Here is the part that saves the most time, and it applies to more sites than people admit.
If nothing on your site ranks in the top twenty, cannibalization is not what is holding you back. Two pages competing for position 58 are not taking a ranking from each other, because neither was close to one. The constraint is authority or indexing, and merging two pages will not change it.
The honest test: look at the positions involved. If both pages sit beyond page three, note the issue in your audit spreadsheet and move on to something that will actually move. Our guide to striking distance keywords covers where the real gains sit.
Cannibalization is a problem for sites with enough authority to rank, where two pages are splitting strength that would otherwise be enough. Below that threshold it is a tidiness issue, not a traffic issue.
How to stop creating it
- Keep a keyword to URL map. One primary query per page, recorded somewhere you will actually look before writing. This single habit prevents most cases.
- Search your own site before writing. A quick check for the target query will show whether you already covered it two years ago.
- Update rather than republish. An annual refresh of the same subject should usually replace the old page, not sit alongside it.
- Noindex thin archives. Tag and category pages compete with the articles they list and rarely deserve to rank.
- Check the results page before publishing to confirm the query really wants a separate page rather than a section inside one you already have, which is part of reading a SERP properly.
Frequently asked questions
What is keyword cannibalization?
Keyword cannibalization is when two or more pages on the same site compete for one search query, so neither achieves the ranking a single stronger page would have. It usually happens by accident, through topic drift, repeated annual updates, or archive pages competing with the articles they list. Two of your pages appearing for a query is not automatically cannibalization.
How do I check for keyword cannibalization for free?
Use Search Console. Open the Performance report, click a query in the Queries tab to filter by it, then switch to the Pages tab, which shows every page of yours earning impressions for that one query. If two or more appear, compare months to see whether the split is stable or alternating. Repeating this for your top twenty queries catches almost everything worth fixing.
Is keyword cannibalization always bad?
No. A stable split where both pages are useful, or where both rank on page one, is a good outcome rather than a problem. The damaging pattern is Google alternating between your pages across weeks so neither settles, or the weaker page winning while the better one stays hidden. A query with almost no searches is never worth the work.
How do you fix keyword cannibalization?
Merging the weaker page into the stronger one and redirecting it is the best fix most of the time, because it consolidates both content and links permanently. The alternatives are genuinely differentiating the two pages, correcting internal links so your site points at the page you want to win, canonicalising, or noindexing the loser. Change one thing at a time, since these take weeks to settle.
Should a new or low traffic site worry about cannibalization?
Usually not. If both competing pages sit beyond page three, they are not taking a ranking from each other because neither was close to one, and the real constraint is authority or indexing. Note it during your content audit and spend the time on something that will move. Cannibalization matters once a site is strong enough that splitting its strength actually costs a position.
Can category and tag pages cause cannibalization?
Yes, and they are among the most common causes. An archive page listing your articles can rank for the same term as the article itself, and it is almost always the worse result for the searcher. Setting thin archives to noindex removes the competition without touching your real content.