You publish a new article hoping to win another spot on the results page, only to discover the new article is competing with your own older article. Two pages target the same keyword, Google alternates between them, and neither holds a stable ranking. This is called cannibalization — multiple pages on the same site competing for one keyword — and it is far more common than most people expect, especially on sites that publish many articles around the same topic.
What cannibalization really is, and when it is not a problem
The first important distinction is between keyword overlap and intent overlap. Two pages can mention the same keyword yet answer two different needs without any problem: an article on "best running shoes for beginners" and one on "best running shoes for marathons" both touch the running-shoe keyword, but they serve different audiences. True cannibalization happens when two pages answer the same question for the same audience, differing only in title.
Google does not penalise a site merely for having several pages appear for one query. The problem lies elsewhere: when two of your pages take turns appearing, clicks, links, and engagement signals get split between them instead of concentrating on one strong page. The result is that neither page is strong enough to hold a high, stable ranking, and you spend effort maintaining two pages that do the same job.
How to find pages competing with each other
The most reliable detection starts with search data: filter by a specific query and see how many of your addresses receive impressions for it. If two addresses both receive meaningful impressions for the same query, they are candidates worth examining. Do not conclude from numbers alone, though: open both pages, read them side by side, and ask whether they genuinely serve the same need.
Another sign sits in ranking history: for one keyword, the displayed page keeps switching between two addresses over many weeks. That is a sign Google itself has not decided which page fits best, and that hesitation usually reflects two pages that are too similar in intent. When tracking rankings, watch not only the position but also which address holds it; a keyword "stable at position 5" that keeps changing pages is a keyword with a problem.
On Keyword.help, you track rankings by country and device down to the Top 30, so page-switching can also differ by device: page A may hold the desktop ranking while page B holds the mobile one. Note such switching cases during tracking, because they form the clearest candidate list for your next content review.
Choose the primary page first, then choose the fix
Before changing anything, choose one primary page for each contested keyword. The primary page should be the best intent match, the most complete content, and the one with the best results in traffic or conversions. Every later decision revolves around it: the other pages will be merged into it, differentiated from it, or removed from the ranking contest another way.
The most common fix is merge and redirect: take the best parts of the weaker page into the primary page, then permanently redirect the weaker page to it. This fits when two pages share an intent and you only need one page. The second fix is differentiation: rewrite the weaker page's title and content to target a clearly different keyword and intent, which fits when the pages actually serve different needs but were initially presented too similarly.
Two further fixes suit special cases. A canonical tag fits when you need to keep both pages for users — near-identical variants, for example — but want to tell Google which is the main version. A noindex tag fits thin utility pages that do not need to appear in search results but still need to exist for site users. Choosing the wrong fix also causes harm: canonicalising two genuinely different pages, or noindexing a page that brings good traffic, can lose rankings without solving anything.
Prevent cannibalization at the planning stage
The most effective prevention is not in fixing but in planning: before writing a new article, check whether your site already has a page targeting the same keyword and intent. A simple table of keyword, intent, and primary-page address prevents most disputes from the start. This rule matters most for teams with several writers, because each writer can unknowingly duplicate another's topic.
Periodic review is also needed, because cannibalization can creep in gradually: an old article updated with a new angle, a category page starting to rank for an article's keyword, or a new landing page accidentally targeting an old keyword. Each quarter, spend one session reviewing your most important keywords for any new page muscling in. Catching it early makes the fix far lighter than merging two pages that have competed for a year.
Common mistakes when fixing it
The first mistake is merging pages just because of surface keyword overlap, without reading the content carefully. A page doing well for its own distinct intent can be deleted by mistake, along with its traffic and accumulated link value. The second mistake is fixing everything at once: merging ten page pairs in one week leaves you unable to tell which pair improved and which got worse. Handle one pair at a time, wait a few weeks to see rankings settle, then move to the next pair.
The third mistake is forgetting internal links after a merge. If other articles on your site still link to the old redirected address, users and Google must pass through an unnecessary intermediate step. After each merge, update important internal links to point straight at the primary page. If you are reviewing content and want to watch how the primary page's rankings change after a merge, you can start at the sign-up page and add that keyword to a tracking project.
Conclusion
Cannibalization is not a penalty; it is a signal that your site is splitting its own strength. Detect it with query data combined with reading both pages directly, choose one primary page per keyword, then pick the right fix: merge and redirect, differentiate, canonical, or noindex. Most importantly, prevent it from the start with a keyword-to-primary-page table. For more rank-tracking methods, see the Keyword.help blog and start your review with your most important keyword group.