Keyword Cannibalization: Find and Fix Competing Pages

Keyword cannibalization is when two or more pages on your own site target the same search intent, so they compete with each other instead of one clear page answering the question. It is an SEO term, not a Google policy, and Google has said that several pages from one site ranking for a query isn't automatically a problem. It becomes a problem when the pages are near-copies, because then it splits your signals and makes a site look padded, which matters for AdSense review too.

What cannibalization looks like

A food blog that has published these over three years:

• /best-sourdough-starter-recipe/
• /how-to-make-sourdough-starter/
• /sourdough-starter-for-beginners/
• /easy-sourdough-starter/

Four pages, one intent: "how do I make a sourdough starter?" Each has 700 words, the same flour-and-water ratio, and slightly different intros. None of them is clearly the best one. In Search Console, impressions for "sourdough starter" are spread across all four, and positions keep swapping.

Compare that with:

• /sourdough-starter/ (how to make one)
• /sourdough-starter-not-bubbling/ (troubleshooting)
• /rye-vs-wheat-starter/ (comparison)

Same topic, three different intents. That's topical depth, not cannibalization.

Why it matters for AdSense, not just rankings

Google's AdSense policies don't mention cannibalization by name. But overlapping pages tend to trigger things that are named:

• Low value content. If a reviewer opens three articles and finds the same content reworded, the site looks like it's padding page count. Google's Publisher Policies call out content that is "rewriting of content from other sources without adding value" (answer/11190248); rewriting your own pages over and over gives a similar impression.
• Scaled content patterns. Dozens of near-identical "best X for Y" pages are a common signature of programmatic SEO done badly, and the spam policies describe "scaled content abuse" as many pages generated "for the primary purpose of manipulating search rankings and not helping users" (spam policies).
• Doorway patterns. City-by-city or keyword-by-keyword variations that funnel to the same answer.

The duplicate content guide covers the copy-paste end of this spectrum. Cannibalization is the subtler middle: pages that aren't identical but don't need to be separate.

How to find it

1. Titles and H1s

Export all your URLs with titles. Sort alphabetically and look for clusters with the same head term. Near-duplicates jump out quickly:

How to Make Sourdough Starter (Easy Recipe)
How to Make a Sourdough Starter From Scratch
Sourdough Starter Recipe: How to Make It

2. Search Console queries

In Search Console's Performance report, filter by a query and switch to the Pages tab. If three or more URLs get impressions for the same core query, and none clearly dominates, you've likely found a cluster.

3. URL patterns

Slugs that differ only by modifiers (best-, easy-, -guide, -2024, -2025) are a common sign. Yearly republished posts ("best laptops 2024", "best laptops 2025", "best laptops 2026") are a frequent source.

4. Content similarity

Tools that compare text shingles can show which pages share 50%+ of their sentences. Approvalens computes title, H1 and body similarity across the pages it crawls and groups likely competing pages.

How to fix it

There are four options. Pick per cluster.

Situation	Fix	Notes
Pages are near-copies, one is clearly stronger	Merge and 301	Move unique bits into the strongest page, redirect the rest
Pages are similar but each has unique value	Combine into one better page	Write the definitive version, redirect all old URLs to it
Pages serve different intents but overlap in titles	Refocus	Rewrite titles/H1s and intros to make the different intent obvious
Variant must stay (e.g. print version, filter page)	Canonical	rel="canonical" to the main page

Merging with 301 redirects

Google's redirect documentation recommends permanent server-side redirects when a page has moved for good. In nginx:

location = /easy-sourdough-starter/ { return 301 /sourdough-starter/; }
location = /sourdough-starter-for-beginners/ { return 301 /sourdough-starter/; }

Then update internal links to point directly at the surviving URL, and remove the old URLs from your sitemap.

Using canonical tags

Canonicals tell Google which URL you prefer when pages are duplicates or very close. Google's guide to consolidating duplicate URLs treats rel="canonical" as a strong hint, not a directive:

<link rel="canonical" href="https://example.com/sourdough-starter/">

Canonicals work best for technical duplicates (parameters, print views, sorted lists). For two editorial articles saying the same thing, merging is cleaner; a reader landing on either still sees thin, repeated content.

Refocusing

Sometimes the pages are legitimately different but the titles hide it. Changing:

• "Sourdough Starter Guide" → "Sourdough Starter Not Rising? 7 Causes and Fixes"
• "Sourdough Starter Recipe" → "How to Make a Sourdough Starter in 7 Days"

makes the intent obvious to readers, and the pages stop looking like duplicates.

Prevent it in the first place

• Keep a simple content map. One row per target intent, with the URL that owns it. Before writing, check the map.
• Update instead of republishing. Refresh "Best laptops" each year at the same URL rather than creating -2027. Change the date only when the content substantially changes; Google's guidance asks whether you're "changing the date of pages to make them seem fresh when the content has not substantially changed".
• Be careful with tags. Tag archives that list the same three posts as a category page create thin duplicate archives. Noindex or remove tags with fewer than a handful of posts.
• Watch AI-assisted drafting. Generating "10 variations" of a topic is the fastest route to a cannibalized site.

A quick audit routine

1. List all URLs and titles.
2. Group by core topic.
3. For each group with 2+ pages, ask: does each page answer a different question?
4. Merge, refocus or canonicalize where the answer is no.
5. Fix internal links and the sitemap.
6. Re-check in four to six weeks in Search Console.

Our methodology explains how Approvalens scores duplicate and overlapping pages.

Find your competing pages

A free scan groups pages with similar titles, headings and body text so you can see clusters before a reviewer does.

FAQ

Is keyword cannibalization a Google penalty?

No. It is not a policy or a penalty. Google has said that multiple pages ranking for one query isn't inherently bad. The issue is split signals and, for AdSense, a site that looks padded with near-duplicates.

Should I delete or redirect overlapping pages?

Redirect (301) when the old URL has links or traffic, and point it to the page that now covers the topic. Delete with a 404 or 410 only when nothing of value points to it.

Is a canonical tag enough to fix cannibalization?

For technical duplicates, often yes. For separate articles that say the same thing, merging is better, because readers and reviewers still see the repeated content.

How similar do two pages need to be to count?

There's no official threshold. If a reader would be satisfied by either page and wouldn't need both, they're competing.