What the Content Cannibalization Comparator does
This comparator finds pages on one site that compete for the same searches, measures how much they overlap - by the queries they share in a Search Console export, by how similar their copy is, or both - and recommends for each pair whether to merge them, canonicalise one to the other, differentiate their intent or leave them alone.
Every recommendation comes with its evidence: the shared queries, each page's share of impressions and clicks, and the terms the copy has in common. The thresholds behind the verdicts are stated and adjustable, because two pages ranking for one query is sometimes useful coverage rather than a problem.
How to use it
- In Search Console, open Performance, add the Page dimension to the query report (or export via the API or Looker Studio) so each row has page, query, clicks, impressions and position. Drop the CSV or paste it. Common header names such as Landing page, Top queries and Average position are recognised.
- Optionally paste the main copy of two to ten pages, separated by a line of three dashes, each starting with its URL. The copy is compared on its own, or joined to the query data for the same URLs.
- Adjust the thresholds if needed: by default a page must hold at least 10% of a query's impressions to count as competing, and queries under 10 impressions are ignored.
- Select Compare. Read the page pairs from the top: canonical and merge candidates first, then pairs to differentiate. Check the contested queries table before acting on any pair.
Reading the results
Query overlap is weighted by impressions: for every query both pages compete on, the smaller of their two shares is counted, then divided by all the impressions of either page's queries. 50% or more means most of the pair's search demand is contested.
Leader's clicks is the share of clicks on shared queries that goes to the stronger page. A clear leader (70% or more) makes a merge simple: keep the leader and redirect the other. A split means neither page is winning, which is often the real cost of cannibalisation.
Copy similarity is the cosine similarity of the pages' word frequencies with common words removed. Above 80% the pages are near-duplicates; around 50% they cover the same ground in different words.
A canonical recommendation appears when two URLs are the same page apart from parameters, protocol, www, case or a trailing slash.
Worked example: the running shoes export
The example export has 16 rows across five URLs. /running-shoes and /running-shoes?sort=price both appear for 'cheap running shoes' - the same page with a sort parameter - so the pair is marked canonical.
The buying guide /blog/best-running-shoes and the category page /running-shoes compete on four queries. Their smaller shares add up to 23,200 impressions out of 73,340 across all their queries, a weighted overlap of 31.6%, and the guide takes 705 of the 1,155 clicks on those queries (61%). That is 'differentiate': make the guide clearly editorial (comparisons, testing) and the category page clearly transactional, and link them to each other.
The trail-versus-road article and the trail shoes category share one query at 23.2% overlap, with the category taking 88% of its clicks - also differentiate, and close to the point where it would be fine to leave them alone.
Formulas and scoring rules
- Competing page
page impressions for query / all impressions for query >= minimum share (default 10%)- Weighted query overlap
sum over shared queries of min(share A, share B) x query impressions / sum of impressions of every query either page appears for- Query Jaccard
shared queries / queries of A or BUnweighted; reported in the CSV.- Copy similarity
cosine = sum(tfA x tfB) / sqrt(sum tfA^2 x sum tfB^2)Words of three letters or more, common words removed, simple plural folding.- Recommendation
same URL -> canonical; overlap >= 50% or copy >= 80% -> merge (leader >= 70% of clicks) or merge or canonical; overlap >= 20% or copy >= 50% -> differentiate; otherwise keepA2Z heuristics, not a Google rule.
Limitations: what the result does not prove
- Search Console anonymises rare queries and samples large exports, so low-volume overlap can be missing.
- Overlap is measured, intent is inferred. Two pages can share queries because they serve genuinely different needs well; read the queries before merging anything.
- Copy similarity ignores structure, images and meaning; synonyms count as different words.
- Merging and redirecting can lose rankings in the short term. Nothing here predicts traffic after a change.
Privacy: where your data goes
Everything you paste, type or drop is processed in this browser tab. It is not uploaded, logged, stored or sent to analytics. Session recording and tag-manager scripts are switched off on this page.
Standards and sources
- Google - Consolidate duplicate URLs - checked 19 Sep 2026
- Search Console Help - Performance report (Search results)
- Google Search Central - Redirects and Google Search
Frequently asked questions
How do I find keyword cannibalization in Search Console?
Export the Performance report with both the Query and Page dimensions, so each row is a page and a query. Queries where two or more of your pages each get a real share of impressions are the candidates; this tool lists them and scores every page pair.
Is it always bad when two pages rank for the same query?
No. A guide and a product page can both deserve to appear, and sometimes both do at once. It becomes a problem when the pages split clicks without either ranking well, or when they are near-duplicates. That is why the tool weighs overlap, click split and copy similarity.
Should I merge pages or use a canonical tag?
Merge and 301-redirect when one page can serve the intent of both and the other does not need to exist. Use rel=canonical when both URLs must stay available - parameter, print or campaign versions of the same content - so Google consolidates signals on one.
What does differentiate mean in practice?
Make each page answer a different intent: for example one informational guide and one transactional category page. Rewrite titles, headings and introductions to match, remove duplicated sections, and link the pages to each other with descriptive anchor text.
Can I use it without Search Console data?
Yes. Paste the copy of two to ten pages and the tool compares their text, flagging near-duplicates and heavy overlap. Query data makes the verdicts much stronger, because it shows what Google actually ranks the pages for.
Does the export leave my computer?
No. The CSV and the pasted pages are read and analysed in your browser. Nothing is uploaded, and nothing is kept after you close the page unless you download the report.
Last reviewed by the A2Z.Tools team against the sources listed above.
