cannibalization check is the process of finding pages on your own site that compete for the same search result, then deciding what to do about each pair. Two pages targeting one intent force a choice between them, and the usual outcome is that neither performs as well as a single strong page would have. The check is unglamorous work, and it consistently produces faster gains than publishing anything new.
The direct answer to the question most people have: a cannibalization check compares the intent behind every page you have, groups the ones that share an intent, and then assigns each pair one of seven outcomes. Only one of those outcomes involves writing something new. The other six are about merging, consolidating, redirecting, or improving pages you already own.
What is keyword cannibalization?
Keyword cannibalization (your own pages competing for one search result) happens when two or more pages on the same site target the same search intent. Search engines have to pick one page to show, and that choice often does not match the page you wanted to rank.
The effect is easy to miss because nothing breaks. Both pages remain published. Neither gets an error. What quietly happens is that signals split between them. Internal links point to two places. Other sites link to whichever one they found first. Impressions scatter across both, and a page that might have ranked well ends up sitting in a lower position than it would with the same content and no competitor beside it.
A cannibalization check is the audit that finds these situations. It is not the same as looking for duplicate content, because cannibalization is about intent rather than wording. Two pages can use completely different words and still compete, provided they promise the same thing to the reader. Within a wider content gap analysis, this is the group of findings that call for fixing existing pages rather than adding new ones.
Why does cannibalization happen so often?
It almost never happens by accident. It happens through four ordinary behaviours.
Publishing on topics already covered. Someone writes an article in month three, and a different person writes a near-equivalent article in month eleven because the internal list was never checked. The two pieces may even link to each other.
Splitting one subject into many pages. A series that grows page by page until five pages each cover a fragment of one subject. Individually they look reasonable. Together they compete.
Updating without consolidating. An older page gets refreshed while a newer one covers the same ground. Now both are current and both target the same query.
Targeting a phrase on a page that already targets a nearby phrase. Page titles drift toward whatever is trending, and two pages end up aiming at the same search without anyone deciding it.
There is a fifth cause worth naming, which is a series of thin pages published quickly. Sites that publish daily or weekly without checking coverage accumulate overlap faster than they can manage it. Google’s spam policies cover scaled content abuse in detail, and that page is worth reading before your publishing cadence becomes a cannibalization check you run constantly instead of once a quarter.
None of these causes require anyone to make a bad decision. Each one is a reasonable local choice that becomes a problem at the site level. That is why a cannibalization check belongs in the publishing process rather than in a crisis response. If your team writes article by article without ever seeing the full inventory, overlap is not a risk. It is the default outcome.
There is also a timing effect that makes the problem worse. Overlap that forms early is cheap to fix, because the losing page has few signals to lose. Overlap left alone for a year becomes expensive, because both pages have accumulated links and impressions that now have to be moved. The cost of a cannibalization check rises with every month you skip it. Google’s spam policies
What are the signs you need a cannibalization check?
Five symptoms show up reliably.
Your pages swap positions for the same query. One week page A ranks, the next week page B does, and the pattern repeats. This indicates the search engine cannot decide between two pages that look equally relevant.
A page you are proud of ranks below a weaker page of yours. If your better page sits below a thinner one for the same query, the two are competing and the search engine picked wrongly.
Impressions are spread across several URLs for one query. If a single query generates impressions on four of your pages, those pages are dividing the same demand.
Your click-through rate is low for a query where you rank well. If you sit in a strong position and still get few clicks, the result shown to the searcher may be the wrong one of your pages.
You cannot say which page should win. This is the simplest test. If you ask yourself which URL is the target for a given search and you hesitate, a cannibalization check will find something.
The fifth sign is the one worth running first, because it requires no tools at all. Sit down with your last twenty published pages and state the one search each of them is meant to win. Any hesitation marks a candidate. You can run this version of a cannibalization check before you decide whether you need a real one.
How do you run a cannibalization check step by step?
Seven steps, and the order matters.
Step 1. Build your page inventory. Every published URL with its title, its focus topic, and its publish date. If you have fewer than a few hundred pages, a spreadsheet is enough. The inventory is the foundation, because a cannibalization check can only compare things you have listed.
Step 2. Write the intent for each page. One line per page. Not the keyword, the intent. “Help someone choose a first mechanical keyboard” is an intent. “mechanical keyboard” is not.
Step 3. Group pages by intent. Sort your inventory so that similar intents sit next to each other. Pages sharing an intent are candidates.
Step 4. Confirm with your own data. Where you have performance data, check whether the candidate pages generate impressions for the same queries. This turns a suspicion into evidence, and evidence is what separates a cannibalization check from a guess. Google Search Console reports which queries each URL appeared for, which is enough to confirm most candidate pairs without any paid software. Google Search Console shows queries at the page level, so you can see two of your URLs sharing one query.
Step 5. Assign one of the seven verdicts. Covered in the next section. Every candidate pair gets exactly one.
Step 6. Check internal links. When pages merge or redirect, incoming internal links have to move. A redirect with no updated internal links loses most of its value.
Step 7. Log the decision and recheck later. Record what you decided and when. Overlap tends to return if the publishing process that created it does not change.
Seven steps sound like a lot until you run them once. Most sites find their candidate groups during step three and confirm them during step four. The remaining steps are mechanical. A cannibalization check on two hundred pages is an afternoon of work, and it usually turns up a handful of genuine conflicts rather than dozens.
What are the 7 outcomes of a cannibalization check?
Every candidate gets exactly one outcome. Choosing deliberately is the whole point.
- New content. Nothing on your site serves this intent. This is the only outcome that means writing a new page.
- Refresh. One page serves the intent but is outdated. Update it rather than adding a second.
- Expand. One page serves the intent but is too thin to be competitive. Add the missing depth.
- Merge. Two or three weak pages cover this between them. Combine the useful parts into one.
- Consolidate. Several pages overlap heavily and none of them justifies its own existence. Reduce to one canonical page.
- Redirect. A page serves no purpose of its own. Point it at the page that does.
- Do not create. The idea is attractive but would only compete with existing work. The most underused verdict.
The distinction between merge and consolidate is worth holding onto. Merge combines pages that each had something useful. Consolidate removes pages that only existed because publishing was easier than planning.
And the distinction between refresh and expand comes down to whether the content is wrong or merely incomplete. Wrong content gets refreshed. Right but thin content gets expanded.
Two of these outcomes are worth explaining further, because they are where most cannibalization check projects go wrong. Consolidate is the verdict people choose when they are tired of the overlap and want the fastest exit. It works when the pages genuinely have nothing unique to offer, and it fails when they do. Check each page for a section, example, or answer that appears nowhere else before you consolidate it. If you find one, merge instead.
Do not create is the verdict that saves the most work over a year. The moment a writer proposes a topic that overlaps an existing page, the cheapest correct answer is often to improve the page you already have. A cannibalization check that ends with three new articles planned has probably not solved anything.
How do you decide which page should win?
Once you have settled on merge, consolidate, or redirect, one page has to be the survivor. Four factors decide it.
Which page has the strongest signals? The one with the most external links, the most internal links, and the best current average position. Preserving an established URL is usually easier than rebuilding authority on a new one.
Which page is better on intent? If one page clearly matches what the searcher wants and the other misses, intent should outweigh raw link strength. A strongly linked page aimed at the wrong reader is still aimed at the wrong reader.
Which URL is cleaner? If both pages are otherwise equivalent, keep the one with the clearer, shorter slug. This is a tiebreaker, not a primary consideration.
Which page do you want to maintain? Practical concerns matter. A page that sits inside a topic cluster you are actively building is a better survivor than a page with no siblings.
Write the answer down before you start merging. Making the decision after the content is combined leads to the merged page inheriting the worst structure of both.
How do you fix overlap without losing what you had?
The mechanics are straightforward and the risk sits almost entirely in the internal links.
Redirect the losing URLs to the survivor. A permanent redirect passes the established signal to the surviving page. Do this after the surviving page has absorbed the useful content, not before, so readers landing from search find a complete page.
Update every internal link. Find each internal link pointing at a redirected URL and point it directly at the survivor. Links that pass through a redirect lose some of their strength, and links that point at a live URL that no longer makes sense damage the reader’s experience.
Absorb the unique parts first. If the losing page answered a question the survivor does not, add that section before redirecting. The merge is where value gets created, and skipping it means you removed a page without replacing what it offered.
Rewrite the survivor’s title and meta. A merged page now covers a slightly wider subject than before. Its title should reflect that, without stretching to cover two unrelated intents.
Recheck after a few weeks. Redirects take time to process. Check whether the surviving page has picked up the impressions the losing pages used to collect. If not, the survivor may need more work rather than less.
One caution. Do not redirect a page to a URL that does not genuinely answer the same question. A redirect that breaks the promise the searcher followed damages trust with both readers and search engines, and it is worse than leaving the overlap in place while you prepare a proper merge.
How do you stop new cannibalization from forming?
Prevention is a publishing process, not a one-time cleanup.
Run the overlap check before writing. Before any new article gets approved, search your own inventory for the intent. This single step prevents most of the problem, and it takes about two minutes per idea.
Keep one owner per intent. Every intent has exactly one page responsible for it. When a subject grows, the existing page expands rather than spawning a sibling.
Assign a focus keyword at the planning stage. One page, one focus intent. If a draft does not have a clear answer to which search it targets, it is not ready.
Review the inventory quarterly. The same cadence that works for a content gap analysis works here. Overlap accumulates quietly, and a quarterly check finds it while the fix is still cheap. A cannibalization check on a schedule is a maintenance task. A cannibalization check after traffic drops is a repair job, and repairs cost more. The wider method behind that cadence is described in content gap analysis.
What does a finished cannibalization check look like?
A short table, one row per candidate group.
| Field | What it records |
|---|---|
| Intent | The one search these pages are competing for |
| Pages involved | The URLs sitting in the group |
| Evidence | Impressions, positions, or the internal link pattern showing the overlap |
| Verdict | One of the seven outcomes |
| Survivor | The URL that keeps the intent |
| Content to absorb | The unique sections moving into the survivor |
| Links to update | Internal links that need to point at the survivor |
| Date decided | When the change was agreed |
If a row has a verdict but no named survivor, it is not finished. Redirecting toward a page you have not chosen is how sites lose the value of a page without gaining anything.
The table also gives you a record for the next round. Six months later, the date column tells you which decisions have had time to settle and which need another look. A cannibalization check is only as useful as the notes it leaves behind, because the same conflict usually returns if nobody remembers how it was resolved.
It helps to keep this table next to the page inventory you built in step one. The inventory records what exists, and the table records what you decided about it. Together they become the planning document that stops the next batch of articles from recreating the overlap you just cleaned up.
FAQ
What is a cannibalization check in simple terms?
A cannibalization check means finding pages on your own site that target the same search intent, then deciding which page should own that intent and what happens to the others. It is an internal audit rather than a competitor comparison.
How is cannibalization different from duplicate content?
Duplicate content means the same or very similar text appears in more than one place. Cannibalization is about intent, so two pages with completely different wording can still compete if they promise the reader the same thing. You can have one without the other.
How often should you run a cannibalization check?
Quarterly suits most sites, and always run one before a content refresh cycle or a batch of new publishing. Sites publishing weekly should check at the planning stage for every new article, which prevents most of the problem before it exists.
Does merging pages always improve rankings?
No. Merging helps when two pages genuinely compete for one intent, because it concentrates signals that were being split. If the pages served different readers or different needs, merging loses coverage rather than gaining strength.
What happens to traffic when you redirect a page?
Some fluctuation is normal in the first few weeks while the redirect is processed. Redirecting to a page that genuinely answers the same question preserves most of the established signal. Redirecting to an unrelated page will lose it, and that loss is deserved.
Can you fix cannibalization without deleting anything?
Often, yes. Expanding one page to cover what a sibling handled, then redirecting the sibling, resolves the overlap. Deleting content is rarely necessary, and it is the option that removes information readers might still need.
Why do search engines not just pick the better page?
They do pick one, but the signals they use are split between your competing pages. Neither page carries the consolidated authority a single page would have. Consolidating gives the search engine one clear answer instead of two partial ones.
Does a cannibalization check matter for a small site?
It matters more for small sites, because they have less authority to divide. A large site can often afford overlap that a small one cannot. Running the check costs an afternoon and regularly outperforms publishing a new article.
Conclusion
cannibalization check work is the least appealing part of managing a site, and it pays better than almost anything else. You are not creating anything. You are removing internal competition so that the pages you already wrote can do their job.
Start with the hesitation test. List what you published recently and state the one search each page is meant to win. Where you hesitate, you have found a candidate group. Write down the intent, name the survivor before you touch any content, absorb what is worth keeping, redirect what is not, and update every internal link that pointed at the old URL.
Then change the publishing process so the problem stops rebuilding itself. A two-minute overlap check before every new article is the cheapest fix in content work, and it is the one that keeps a cannibalization check from becoming an annual event.