Keyword cannibalization: how to find and fix it

Keyword cannibalization happens when two or more pages on your site compete for the same query, so search engines have to pick between them and your ranking signals get split across both. The result is usually that neither page ranks as well as one strong page would have. You find it by filtering Search Console for a query and checking how many of your URLs collect impressions for it, and you fix it by consolidating, redirecting, re-targeting the intent, or de-optimizing the weaker page. The diagnosis takes twenty minutes. The judgment about which page survives is the real work.

What keyword cannibalization actually is

The name suggests your pages are eating each other. What is really going on is more mundane: Google generally shows one result per site for a given query, so when three of your pages are plausible answers to "keyword clustering", it has to choose. The choice is not always the one you would make, and it can flip from week to week. Meanwhile the links, the internal link equity and the engagement signals that should have piled onto one authoritative URL are spread thin across three mediocre ones.

It is worth being precise, because the term gets thrown at problems it does not describe. Two pages mentioning the same keyword is not cannibalization. Two pages ranking for overlapping long-tail variants is not cannibalization either, and it is often fine. Cannibalization is specifically when pages target the same search intent, so that a single searcher typing a single query would be equally well served by either one. If a reader would be happy landing on either page, you have written the same page twice.

The symptoms worth watching for

• The URL keeps swapping. One week page A ranks for the query, the next week page B does, and the position wobbles around the same mediocre band. That churn is Google failing to pick a clear winner.
• Impressions split across URLs. In Search Console, one query shows meaningful impressions spread over two or three of your pages instead of concentrating on one.
• A strong page stalls. A page you have invested in sits at position 8 to 12 and will not move, while a thinner page on the same topic hovers just behind it.
• The wrong page ranks. Your blog post ranks for a commercial query that your landing page was built to win, so the traffic arrives and never converts.

That last one is the expensive version. It is not a ranking problem, it is a revenue problem, and it is easy to miss because the traffic graph looks fine.

How to find keyword cannibalization

Three checks, in increasing order of effort. Run the first two before you touch anything.

1. Google Search Console, filtered by query

This is the authoritative source, because it shows what Google actually does with your site rather than what a third-party tool estimates. Open the Performance report, filter to a query you care about, then switch to the Pages tab. If a single query is pulling impressions across two or more of your URLs, you have a candidate. Sort by impressions and work down the list. Queries where the split is roughly even are the worst offenders, because that is Google genuinely unable to decide.

2. The site search

Type site:yourdomain.com "your keyword" into Google. You will get every page on your site that Google considers relevant to that phrase, ranked roughly by how relevant it thinks each one is. If the top two results are both yours and both look like they are trying to answer the same question, that is your confirmation. It takes ten seconds and it is free.

3. A crawl of your own titles and H1s

Export every page title and H1 on the site and sort them alphabetically. Duplicates and near-duplicates jump out immediately. This catches the cannibalization you created on purpose and forgot about: the three "best X software" posts written by three different people in three different quarters. Checking keyword difficulty on the shared term at the same time tells you whether the consolidated page has any hope of ranking once you have built it.

The four fixes, and when to use each

Not every case gets the same treatment. Pick the fix that matches what the two pages actually are.

Consolidate and redirect

This is the default and usually the right answer. Take the two competing pages, decide which URL is the keeper (normally the one with more links, more traffic history, or the better URL), merge the genuinely useful material from the loser into it, and 301 redirect the loser to the keeper. The redirect passes the link signals across so the surviving page ends up stronger than either was alone. Do not skip the merge step and just delete the weaker page: whatever made it rank at all is worth keeping.

Re-target the intent

Sometimes both pages deserve to exist and you simply pointed them at the same query by accident. A guide about "what is keyword clustering" and a product page for a clustering tool serve genuinely different searchers, and they only collide because both were optimized for the bare term. The fix is to pull them apart: point the guide at the informational phrasing, point the landing page at the commercial phrasing, and rewrite the titles, H1s and opening paragraphs so each one is unmistakably aimed at its own reader. Search intent is the axis you separate on, not word count.

De-optimize the weaker page

When a page is valuable for its own topic but keeps drifting into a query it was never meant to win, you do not need to delete it. Strip the target keyword out of its title, H1 and headings, keep it in the body only where it reads naturally, and add an internal link from that page up to the page that should rank. You are telling both Google and your readers which URL is the real answer.

Canonicalize, carefully

If two pages must both stay live and reachable at their own URLs but are genuinely near-duplicates, a canonical tag on the weaker one pointing at the stronger one consolidates the signals without a redirect. Use this sparingly. A canonical is a hint, not a directive, and Google will ignore it if the pages look too different. It is a tool for genuine duplicates, not a way to avoid making an editorial decision.

Why cannibalization keeps happening

Almost every case traces back to the same root cause: somebody planned content one keyword at a time. You pick a keyword, you write a post. Three months later somebody picks a keyword that is functionally the same question worded differently, and writes another post. Nobody notices, because nobody was holding the whole map. Do that for two years and you have forty posts and a dozen quiet collisions.

The structural fix is to plan in clusters rather than keywords. Group every keyword by the intent behind it, and let one cluster own one page. When a new keyword arrives, it either joins an existing cluster (so it becomes a section on a page you already have) or it forms a new one (so it earns a new page). It cannot quietly become a second page targeting an intent you already covered, because the map will not let it. That is the whole argument for building topic clusters, and it is why the keyword clustering tool groups a keyword list before it hands you anything to write.

What to do after you fix it

Two things. First, be patient: after a consolidation Google needs to recrawl both URLs, follow the redirect and re-evaluate the merged page. Movement typically shows up over a few weeks, not a few days, and it is normal to see a brief dip before the recovery. Do not panic and revert on day four.

Second, remember what the page is for. A consolidated page that finally reaches position 3 is only worth the work if the visitor who lands there does something. This is a good moment to look hard at the copy, layout and call to action on the page you just made your winner, because you are about to send it every visitor those two old URLs used to split between them. Ranking was the hard part. Wasting the traffic afterwards would be worse.

The short version

Cannibalization is a planning failure that shows up as a ranking symptom. Find it by filtering Search Console by query and watching for impressions spread across multiple URLs, confirm it with a site search, then consolidate the pages that answer the same intent and separate the ones that only looked alike. Then change how you plan, so it stops happening: cluster the keywords first, and give each intent exactly one page. If you want the whole route from a seed keyword to a clustered plan, the guide on how to do keyword research for SEO covers it end to end, and the SEO content plan is what the finished map looks like.