You commit to publishing more, you follow through, and six months later rankings for your best page have slipped. No competitor overtook you. You did it to yourself, and the reason it went unnoticed for six months is that nothing appeared to break.
What cannibalisation actually is
Two or more pages on your site targeting the same search intent. Note the wording — the same intent, not the same keyword. Two pages can share no exact phrasing and still be answering an identical question, and search engines are matching intent, not strings.
When it happens, a search engine has to pick one page to show. It frequently does not pick the one you would have. Meanwhile your internal links, your external links and your engagement signals are being divided between the candidates instead of accumulating behind one.
Why it is hard to notice
There is no error state. No page 404s, no warning appears in any tool by default. Your strongest page just drifts down the results while total site traffic looks broadly flat, because the new pages are picking up something.
By the time it is obvious in a traffic graph, you usually have several months of content built on the same faulty assumption.
Cannibalisation does not announce itself. It just quietly caps how well any single page can do.
How to find it
Three checks, in ascending order of effort:
- In Search Console, filter to a single query and look at the pages receiving impressions for it. More than one URL trading places over time is the clearest signal you will get.
- Run a site-restricted search for the query and see which of your pages the engine actually surfaces, and in what order.
- Export your page titles and sort them. Near-duplicate titles are the cheapest possible tell, and they turn up more often than anyone expects.
Deciding which page wins
Once you have found a conflict, one page has to be chosen. Pick on evidence rather than sentiment.
The strongest candidate is usually the page with the most earned links, or the best conversion rate, or the closest match to the actual search intent. It is very often not the newest page, which is inconvenient, because the newest page is usually the one somebody just finished writing.
The four fixes
Which one applies depends on why the pages exist:
- Consolidate. Merge the useful content into the winner and 301 redirect the weaker URL. Best when both pages are genuinely trying to do the same job.
- Differentiate. Rewrite one page to serve a genuinely distinct intent. Only honest if such an intent exists — do not invent one to avoid deleting something.
- Canonicalise. When both pages must remain for users, point the canonical at the one that should rank.
- De-optimise. Strip the competing signals from the lesser page: retitle it, adjust the headings, and re-point internal links at the winner.
Preventing it
Cannibalisation is a planning failure, not a writing failure. The prevention is a keyword-to-page map that exists before anything gets commissioned: one intent, one page, written down.
Then make checking that map a step in the brief. It takes two minutes and removes the entire class of problem.
Publishing more is only a strategy if every new page has a job that no existing page is already doing. Without that check, volume stops being growth and becomes interference.
