Keyword cannibalisation:when your pages compete with each other

Keyword cannibalisation means that two or more pages of the same website target the same search query. Google then picks one itself and that pick rarely lands on the page you intended; the other one sinks. You recognise it by titles that largely share the same words and by a position that alternates between two URLs in Search Console. The fix is giving every page one search query of its own and enforcing that with the title, the headings and the internal links.

Why it costs you positions

Google rarely shows two pages of the same website for one search query. When two of your pages claim the same query, the algorithm picks one and you cannot steer that pick. In practice the page with the most link value usually wins and that is almost always the home page, not the landing page you built specially for it.

The damage is double. Clicks and link value spread across two pages that each become half as strong and the position fluctuates because Google alternates between the two. Two half pages rarely beat one whole page of a competitor.

For AI answer engines it weighs even heavier. A composed answer has room for two or three sources, not ten, as we explained in the chapter on GEO. A website that spreads its authority on a topic across two half pages reaches that bar with neither.

How it arises without anyone noticing

Nobody builds two pages for the same search query on purpose. It creeps in while a site grows: a blog article on a topic that already has a service page, a new landing page without anyone checking the old one, or an overview page carrying the same words as the detail pages beneath it.

We found it on our own site. Our home page and our landing page for the term software advisory firm carried almost the same title; Google chose the home page and the landing page that was built precisely for that term sat there for nothing. And in our English and French versions, pages collided that were neatly separated in Dutch: three different Dutch titles had converged on the same wording in translation. Translations deserve their own check.

How to establish it

Three checks, all three free. Do them for every main term you want to be found on.

Check How Cannibalisation when
Site search Search Google for site:yourdomain.be "your term" Two pages both carry the term in their title
Search Console Performance, filter on the query, open the Pages tab Two URLs get impressions and trade places in position
Titles side by side Strip the brand name and stop words from two titles and compare what remains The majority of the content words is shared

The third check is the strictest and also finds the cases that do no damage yet but will. Titles like Independent ERP advice and ERP advice without sales interest look different, but after stripping the trimmings what remains is ERP advice twice.

The fixed division of roles: buying or understanding

Behind almost every topic sit two kinds of search queries. Whoever wants to buy searches for have built, advice or firm. Whoever wants to understand searches for what does it cost or how does it work. Those are different visitors in different stages and they each deserve their own page.

Fix that division: the service page wins the commercial query, the blog article or guide wins the informational one. They then do not compete but reinforce each other, because the guide links to the service for whoever is ready to choose and the service links to the guide for whoever wants to understand first. And a tool such as a free check or calculator never uses the main term of the service page it belongs to.

Four ways to fix it

In order of how often they are the right choice:

  1. Rewrite the weakest one. Give one of the two pages a search query of its own in title, H1 and description. Usually this suffices: the pages existed for good reasons, only their words collided.
  2. Merge and redirect. If both pages are half, make one strong page and put a 301 redirect on the old URL. Never delete a page without a redirect, because then you lose the link value it built.
  3. Aim your internal links. The anchor text of a link tells Google what the destination is about. A link to the service page business analysis should therefore read business analysis and not request a call. When all your anchor texts for a term point to the same page, Google knows which one you mean.
  4. Canonical, only for near-duplicates. If two pages are virtually identical, for instance a print version or a filter variant, point out the main version with a canonical link. For two substantively different pages this is the wrong tool.

Preventing it: one search query per page, before you write

The cheapest fix is the collision that never happens. Make it an editorial rule: whoever adds a page or article first names the search query it should win and then checks whether an existing page already claims it. That takes two minutes with the site search above.

If your site has multiple languages, check per language. And whoever has a developer in house can have it guarded: a script that compares all page titles per language on every change and raises an alarm when two titles share too many content words. Since we started doing that ourselves, a collision can no longer slip online quietly.

How to then write those titles and headings well is in the previous chapter. And whether AI answer engines effectively cite your strongest page is what you measure with the citation test later in this series.

Would you rather not do this yourself: SEMANU carries out the full SEO and GEO audit for 1,450 euro, report and working session included. What it covers is on the guide page.

Frequently asked questions

What is keyword cannibalisation?

Keyword cannibalisation means that two or more pages of the same website target the same search query. Google then picks one itself; the other sinks and the position alternates between both URLs.

How do I establish keyword cannibalisation?

Three free checks: a site search per main term, the Pages tab under Performance in Search Console with a filter on the query, and laying two titles side by side after stripping brand name and stop words. If they share most content words, they claim the same query.

How do I check keyword cannibalisation for free in Search Console?

Open Performance in Search Console, filter on the query and open the Pages tab. If two URLs appear in turn with a position that alternates week to week, that is the signal. Then compare the titles after stripping brand name and stop words.

How do I fix keyword cannibalisation?

Rewrite the weakest page towards a query of its own, or merge both with a 301 redirect. Also aim the internal anchor texts at the right page and use a canonical only for near-duplicates.

Measurability and findability belong in your software strategy

Measurability and findability belong in your strategic advice. If you would rather not do this yourself, we take it up in the software strategy: what you measure, which threshold justifies an action and who follows it up.