Learn
What keyword cannibalisation is, and how to fix it.
It is what happens when two or more of your own pages try to rank for the same search. Instead of one strong page, you give Google several weaker ones and ask it to choose. It often chooses the wrong one, or none of them.
What it looks like from the outside
You wrote a service page for end of tenancy cleaning. Over the years you also wrote a blog post about end of tenancy cleaning, a checklist for end of tenancy cleaning, and a page about end of tenancy cleaning in Stockport that says the same things with the town added. Each one is fine on its own. Together, they compete.
Google usually shows one page from a site for a given search. When four of yours are plausible answers, it has to guess which one you mean. Sometimes it shows the blog post instead of the page with your prices and booking button. Sometimes it swaps between them week to week, so none of them builds a steady position. Sometimes it decides the whole set is repetitive and keeps several of them out of the index.
Why Google does not simply rank all of them
It can, and occasionally it does show two pages from one site for the same search. But it mostly avoids filling a results page with one site's near-identical answers, because that is a worse page for the searcher. So your pages are not only competing with rivals. Each one is also competing with its siblings for the one slot your site is likely to get.
The strength that decides who wins, such as links from other sites, links from your own pages and the history of clicks, is split between the siblings instead of pooled into one. Four pages with a quarter of the strength each usually lose to one rival page with all of it.
There is a second cost that is easy to miss. The page Google picks may not be the one you want customers to land on. An article that explains a service is a fine page, but if it outranks the service page, the searcher who was ready to book lands on a page without a price or a booking button.
How to tell if you have it
Look at the search, not the page
In Search Console, open Performance, filter by a search you care about, then switch to the Pages tab. If three or four of your pages all collect impressions for that one search, they are sharing it.
Watch for the ranking page changing
If the page Google shows for your main search keeps switching between two of your addresses, that is Google failing to decide. A position that jumps around with no change on your side is often this.
Check the refused pages
Pages in the Crawled, currently not indexed bucket that cover the same topic as a page that is indexed are frequently the losers of this contest.
Read your own titles side by side
Export every page title into one list and sort it. Near-identical titles are the quickest map of where the overlap is.
The fix is almost never deleting
The instinct is to delete the extra pages. Deleting throws away whatever those pages had earned: links from other sites, the odd search only they ranked for, and the history Google has with them. The better fix is to decide which page should own the search and fold the others into it.
Merge the useful parts of the weaker pages into the page that should win, then redirect the old addresses to it with a permanent redirect, so anything they had passes to the winner. Where a page genuinely answers a different question, keep it, sharpen it so it only answers that question, and link from it to the main page with words that describe the main page.
Merging two pages without losing what they earned
Pick the address that stays
Usually the page with the most links from other sites and the clearest match to the search. If both are close, keep the one whose address you would want customers to see.
Move the useful parts, not everything
Bring across the paragraphs, examples and questions the other page answered better. Leave out the repetition. The merged page should be better than either original, not longer by default.
Redirect the old address permanently
A 301 redirect sends visitors and Google from the old address to the page that stays, and tells Google the move is permanent. Never leave the old page live with the same content.
Update your own links
Change internal links that pointed at the old address so they point straight at the winner. Redirects work, but a direct link is cleaner and faster.
Write it down and check in a month
Record which address went where and why. After a month, check in Search Console that the search now shows one page, and that its position has settled.
Four decisions, one per page
Keep and lead
The page that best matches what the searcher wants to do. For a service search, that is almost always the service page, not the article.
Merge and redirect
Pages that say the same thing as the leader. Their best paragraphs move into it; their address points to it.
Narrow
Pages that overlap but have a real job of their own. Rewrite so they answer one specific question, and link up to the leader.
Leave alone
Pages that only look similar in the title but serve a different search. Not every overlap is a problem.
The four decisions, worked on four real-looking pages
Take the four end of tenancy pages from the top of this page. The service page keeps and leads, because it has the price and the booking button and matches the search from someone ready to hire. The general blog post merges into it: its best paragraph on what landlords check moves across, and its address redirects.
The checklist narrows. It becomes the answer to a different search, an end of tenancy cleaning checklist that tenants print, and it links up to the service page with the words end of tenancy cleaning. The Stockport page stays only if it can say something true about Stockport, such as the estates the firm covers and how quickly a van gets there; otherwise it merges too. Four competing pages become two that help each other.
Where cannibalisation usually comes from
Almost always from publishing without a map. A blog written week to week, each post chosen because it seemed a good idea at the time, will eventually cover the same ground twice, then five times. So will town pages written from one template with the name changed, and service pages that were split into variations nobody searches separately.
It also comes from agencies that report on how many pages they published. If the measure is volume, overlapping pages are the cheapest way to hit it. That is one reason I date every piece against a real search before it is written: a new page that would compete with an existing one is caught in the plan, not in the index.
What it did on a real site
A Las Vegas injury firm had years of posts that quietly said the same things as each other. When Google tightened how it treats older sites, 4,144 of its pages were refused. I merged the overlapping posts into single stronger pages, one written and reversible decision at a time, and deleted nothing in bulk. By 3 September 2026, Search Console counted 2,549 refused, and the pages that remained were the ones doing the work.
Prevention is simpler than cure. Decide what each page is for before it is written, with one search per page, and cannibalisation mostly never starts. That is what a site structure is for.
Questions people ask about cannibalisation
Is it a penalty
No. Nothing is being punished. Google is choosing between your pages, and the result is a weaker position than a single clear page would earn.
Can a blog post and a service page target the same keyword
They should target different searches. The service page takes the search from someone ready to buy; the article takes a question asked earlier, and links to the service page.
Should I use a canonical tag instead of a redirect
A canonical tag suits pages that must both exist, such as a printable version. If a page has no reason to exist alongside the other, merge and redirect; it is the clearer signal.
How many pages are too many on one topic
There is no number. What matters is whether each page answers a different question. Twenty pages can each have a job; two can be one too many.
Will fixing it make me drop first
Sometimes positions move around for a few weeks while Google recrawls and consolidates. That is why the record of each decision matters: you can tell settling from a real problem.
Think your pages are competing?
Send me the site and the search you care about. I will show you which of your pages are sharing it.