Keyword cannibalization is one of the most over-diagnosed problems in SEO. Run any audit tool across a site of a few hundred pages and it will flag dozens of “cannibalised” keywords, most of which are harmless overlaps. The real version is narrower and more damaging: pages that compete for the same intent, split the signals between them and leave Google choosing a different one each week. Finding that version, and leaving the rest alone, is the whole job.
What keyword cannibalization actually is
Semrush’s guide, “Keyword cannibalization: How to find, fix, and prevent it”, defines it as multiple pages on one site targeting the same keyword or keywords and harming each other’s search visibility. The harm takes three forms:
- The wrong page wins. Search engines may pick one of the pages, and not necessarily the one you would choose.
- Links are split. Internal and external links that could strengthen one page are divided between two or more.
- Everything drops. Rankings for all the competing pages can fall.
The key word in the definition is “harming”. Overlap is normal on any site that covers a subject in depth. Cannibalization is overlap with a cost.
A typical keyword cannibalization example looks like this. A site publishes a guide to choosing a product, then a year later a “best of” list for the same product, with a near-identical title and the same recommendations. Both target the same buyer at the same stage. In Search Console the query that used to send traffic to the guide now shows both URLs, each appearing for a few days before the other replaces it, and the combined clicks are lower than the guide earned alone.
How to check for keyword cannibalization
Start in Search Console, not in a third-party audit. Open the Performance report, filter to a single query, then open the Pages tab. If two or more of your URLs appear, switch the chart to show them over time.
The signal is alternation. Two URLs trading places, with impressions moving from one to the other and back, mean Google cannot decide which page answers the query. A position that wobbles as the ranking URL changes is the clearest evidence you will get. By contrast, two URLs that both hold steady positions are not competing in any sense that matters.
A site search for the topic is a useful second check. Search your own domain for the core phrase and read the titles that come back. If you cannot tell from the titles which page should rank for which query, neither can a reader.
Two pages that both rank is a win. Two pages that take turns is cannibalization.
When it is not a problem
The received wisdom treats any shared keyword as a fault to be fixed. It is not. Two results from one site on the same results page can take more of the page than one result would, and that is a gain, not a loss. Pages that rank for the same broad term but answer different intents, such as a buying guide and a how-to, often belong together.
Merging pages that were both performing can cost you the traffic of the one you remove. Before you fix anything, ask what the pair earns together now, and whether a single page would plausibly earn more. If the answer is unclear, leave it and watch.
How to fix keyword cannibalization
Every fix begins with one decision: designate one preferred page per keyword. Map the competing URLs in a spreadsheet, note which has the better links, engagement and fit to the intent, and choose. Then apply one of five techniques Semrush sets out:
| Technique | Use when | Watch for |
|---|---|---|
| 301 redirect | The weaker page adds nothing the preferred page lacks | Move any unique content across first |
| Canonical tag | Both pages must stay live, for example for users or campaigns | Absolute URL; the target must return 200 and must not be noindexed |
| Differentiate the content | The pages could serve genuinely different intents | Retitle and rewrite so each targets its own query |
| Noindex | The competing page is useful on-site but should not be in search | The page leaves search entirely, so be sure it earns nothing there worth keeping |
| Internal links | The preferred page needs to be clearly signalled | Point descriptive anchors at the preferred page, not the competitor |
A redirect is the strongest and least reversible option. Differentiation is the most work, and often the best result, because it can turn one weak pair into two pages that each rank for something. Our guide to content pruning covers the redirect and merge decisions in more detail.
Merging two articles
If you merge, write the combined page properly rather than pasting one article under the other. Keep the best answer at the top, carry over any section the weaker page did better, then redirect the old URL and update internal links to point at the survivor. Leaving internal links pointing at a redirect works, but it is untidy and easy to fix. Then watch the query in Search Console for a few weeks. One URL holding the position where two used to alternate is the confirmation that the merge did its job.
Preventing it in the first place
Cannibalization is mostly a planning failure. Sites that publish from a keyword list, one page per variant, produce it in bulk: “best X”, “best X for beginners” and “top X” end up as three thin pages chasing one intent. A keyword-to-URL map, kept up to date and checked before every new brief, prevents most of it.
Topic clusters help too. A pillar page for the broad term, with supporting pages each owning a distinct sub-question, gives every page a job; our post on building topical authority sets out the structure. To see how sites that already rank in your niche divide a topic, open the database explorer, filter to your niche and sort by estimated traffic, then run a site search on the leaders for your core phrase and note how many pages each gives it. More strategy guides of this kind are collected in the strategy archive.
The limitation: none of this tells you for certain why Google chose one URL over another. Search Console shows which page appeared, not why, and a rank tracker shows the same thing at a coarser grain. Alternation is strong evidence of competition, not proof, so change one thing at a time and give each change time to settle.
If rankings fell across many pages at once, cannibalization is rarely the explanation. Check the timing against confirmed updates first; our guide to recovering from a core update covers that diagnosis.
Frequently asked questions
Is keyword cannibalization bad?
Only when the pages harm each other’s visibility. Two pages from one site ranking steadily for a query can be a win. The problem is pages alternating so that neither holds a position.
How do I check for keyword cannibalization?
In Search Console’s Performance report, filter to one query and open the Pages tab. Two or more URLs alternating over time is the signal.
How do I fix keyword cannibalization?
Choose one preferred page per keyword, then use a 301 redirect, a canonical tag, differentiated content, noindex on the competing page, or internal links pointing to the preferred page.
The takeaway Keyword cannibalization is alternation, not overlap. Confirm it in Search Console, choose one page per keyword, and pick the least destructive of the five fixes that solves it.
