How to fix keyword cannibalization: find it, then merge or retarget

The idea is easy to overstate. In Ahrefs' guide, "you only have a real cannibalization issue when multiple pages target the same keyword and hurt a site's organic performance." When Ahrefs reviewed a sample of 80 keywords for which its own site ranked with more than one page, it found "only one case that needed action." So the work comes in three steps: find candidates, confirm they are real, then fix the ones that are.

When two pages on one keyword are fine

Pages rank for hundreds of queries each, so some overlap between them is normal. Merging two pages because they share one query can cost both pages their other rankings. Ahrefs' warning: "If we were to merge or delete one of them, we'd likely lose some of our other keyword rankings and see a net drop in traffic."

Different intents are not cannibalization either. Ahrefs ranks for "SEO audit" with a blog article and with a tool landing page, and keeps both: "Consolidating them would make no sense because we would need to choose between catering to searchers who want a tool and those who want a guide." Semrush adds branded queries, where several pages from the brand's own site ranking is expected.

Google allows for it too. Describing its site diversity system, Google says "we generally won't show more than two web page listings from the same site in our top results," and more than two when it is "especially relevant to do so for a particular search."

The symptoms of real cannibalization

Ahrefs' study defines the bad case as "when a search engine constantly exchanges ranking between multiple pages or when multiple pages rank simultaneously for the same keyword but are similar enough to be consolidated." In practice it looks like this:

• The ranking URL keeps swapping. In one Ahrefs example, every new case study it published pushed the previous one out of the results for "seo case studies."
• Two similar pages sit side by side in the middle of the results, and neither climbs.
• The wrong page ranks, such as an outdated post instead of the guide you keep current.
• One query's clicks and impressions are split across two URLs in Search Console.

How to find keyword cannibalization

1. Check queries in Search Console

In the Performance report, open Search results, click a query to filter by it, then switch to the Pages tab. Semrush's rule: if more than one URL earns clicks and impressions for the query, treat it as a candidate.

2. Run a site: search

Search site:yourdomain.com with the topic in quotation marks to see every page Google associates with it. Ahrefs warns that "Google returns every vaguely matching result," so treat the list as candidates to check, not a diagnosis.

3. Look at ranking history

A rank tracker shows which of your URLs ranked for a keyword over time, and swapping lines are the clearest sign there is. Ahrefs' Site Explorer has a "Multiple URLs only" toggle for this, and Semrush's Position Tracking has a Cannibalization tab.

4. Fold it into a content audit

On a small site, overlapping pages stand out once every URL is listed with the query it targets, which is part of any content audit.

Then confirm the intent before changing anything. Read both pages next to the results for the query. If both answer the same question for the same reader, it is cannibalization; if they serve different needs, leave them alone. Reading intent off the results page is covered in search intent.

How to fix keyword cannibalization

First choose the page you want to rank. Ahrefs suggests working with the page Google already ranks highest and redirecting the others there; Semrush adds backlinks and traffic to the comparison. Then pick the fix that fits:

Situation	Fix
Same intent, one page clearly stronger	Merge the weaker page into it and 301 redirect
Same intent, both pages weak	Write one better page at the stronger URL and redirect the rest
The weaker page has its own useful angle	Retarget it at a different, more specific query
Same keyword, different intents	Differentiate both and keep them
Near-identical copies that must stay reachable	Canonical tag pointing at the main version

Merge and redirect

This is the fix for most real cases. Move anything unique from the weaker page into the stronger one, publish, then redirect the weaker URL permanently; Google recommends "a permanent server-side redirect whenever possible" when the URL shown in search results should change. Ahrefs did this with two guides on broken link building, and afterwards "only one gets traffic… but it gets way more traffic than both pages (combined) did beforehand." Finish by updating internal links that pointed at the old URL and removing it from the sitemap, as Semrush advises. When merging beats deleting is covered in content pruning.

Retarget the weaker page

If the weaker page has a reason to exist, give it a different job. Rewrite its title, H1 and opening around a narrower query it can own, usually a more specific long-tail phrase, and cut the sections that repeat the main page.

Differentiate by intent

When a keyword is ambiguous and each page serves one meaning, sharpen both: make the guide more clearly a guide and the product page more clearly a product page. Semrush's advice is to give each page "unique angles, examples, or use cases instead of repeating the same information."

Point internal links at the winner

Link to the page you chose, with anchor text that names its topic. Consistent internal linking is one of the clearer signals of which page is the main one. A canonical tag is only for near-identical pages, such as parameter URLs or a print version, not for two different articles. Ahrefs is blunt about it: "Canonicalization is not a way to fix keyword cannibalization." Near-copies are a duplicate content problem, with their own fixes.

Fixes to avoid

• noindex. It drops the page from the index, so it ranks for nothing at all. Ahrefs calls it "a terrible way to fix cannibalization," and Semrush keeps it for use "as a last resort."
• De-optimizing one page. Removing the keyword or its anchors from the weaker page tends to hurt that page's other rankings too, because, as Ahrefs puts it, "you can't de-optimize a page for just one keyword."
• Deleting without a redirect. The page's links and rankings disappear with it.

Prevention is cheaper than any of these. Keep a keyword map with one primary keyword per URL and check it before writing anything new. Semrush's rule: "If a page already exists, optimize the existing page rather than creating a new one."

How Structura handles this

Prevention is built into campaigns. Each keyword in a campaign's bank gets one post, and Structura keeps a record of the keywords it has already published posts for on your site, so a covered keyword stays covered even when another campaign or a refreshed keyword list proposes it again. A second post under the same keyword is written only for a genuinely different long-tail phrase the research found. And if a finished scheduled post would go live at a web address your site already publishes under, the run stops and you get a notification; posts you request yourself with Generate Now skip that check.

That record covers only what Structura published. A keyword one of your own older articles targets is not in it, and the address check catches an exact clash, not two different URLs on one topic.

For those cases there is detection. On paid plans with Google Search Console connected, Structura re-checks your search data every morning and flags any query where two or more of your pages each take a substantial share of the impressions, naming the competing pages, including ones Structura did not write. The suggested fix is the one above: point the pages at different queries, or consolidate them. Structura does not merge, redirect or edit anything itself; redirects and canonical tags stay with WordPress, your SEO plugin or your front end. The blog SEO software page shows how these checks fit together.

FAQ