Keyword Cannibalization Checker

About the Keyword Cannibalization Checker

Keyword cannibalization is two or more of your own pages showing for the same search, so neither gets the clicks one strong page would. This checker reads query and page rows from Search Console — or a rank tracker’s keyword and URL list — and finds every query where pages compete:

• the split of impressions and clicks between the pages, their positions, and which one leads;
• unstable queries, where Google switches from one page to the other from day to day or between two periods;
• double listings in the top 3, which are usually good and need nothing;
• the same page under several URL variants (www, http, a trailing slash, letter case, parameters).

Competing pages are grouped in pairs, each with a recommendation: keep one, merge the other into it, redirect it, or re-target it at its own searches — with the steps and a redirect list. Everything is read on your device.

Common uses

• Finding why a page that used to rank now swaps places with another page of the same site.
• Cleaning up after years of blogging: old posts, updated copies and tag pages that compete with the article people should land on.
• Planning a content merge or a migration, with the redirect list ready for the server.
• Auditing a client’s site with every competing pair, its numbers and its recommendation, as a PDF or a spreadsheet.

Files it reads

• Search Console API responses (searchanalytics.query with the query and page dimensions; also with date, device or country), as JSON or as a sheet: Query, Page, Clicks, Impressions, CTR, Position — the Search Analytics for Sheets add-on writes this sheet.
• Looker Studio URL impression table (Query, Landing Page, Url Clicks, Impressions …).
• Bulk data export to BigQuery (searchdata_url_impression: url, query, impressions, clicks, sum_position; anonymized queries are counted, not listed; Web results only when other search types are in the file).
• Performance exports filtered to one page each: their Filters sheet names the page, so their queries become that page’s.
• Rank trackers: a keyword, a URL and a position (Semrush, Ahrefs and similar exports, Ahrefs’ previous URL too).

CSV, TSV, Excel and JSON; UTF-8 or UTF-16; Indian and Western digit groups and decimal commas.

How conflicts are found

• URLs that differ only in www, http or https, a trailing slash, letter case, an index file, tracking parameters or a #fragment are one page: they are counted together and listed under URL variants.
• A query is a conflict when two or more pages each get at least 10% of its impressions and at least 10 impressions. Search Console counts one impression for each of your pages shown in a search when data is grouped by page (how impressions are counted), so the shares say how often each page was shown. In a rank tracker’s file, pages compete when both rank in the top 20.
• Double listing: both pages average a top-3 position. Unstable: the page with the most clicks changes at least twice from day to day, the second page leading at least a fifth of the days, or another page led in the earlier period.
• Priority: the impressions that went to pages other than the leading one, × 1.25 when unstable and × 0.1 for a double listing.

How keep, merge, redirect and re-target are decided

• Keep the page with more clicks for the shared queries — unless it is the home page or a listing page (tag, category, page 2, site search) competing with a page made for the query; then the page made for the query is kept.
• Redirect the other page when 60% or more of its own impressions come from these queries and it gets under a quarter of the kept page’s clicks; merge it first when it gets more. Google uses a permanent redirect as a signal that its target should be the canonical URL (redirects and Google Search).
• Re-target it when most of its impressions come from other queries: it has an audience of its own, so its title, main heading and links move to those searches.
• A home page or a listing page that competes links to the page made for the query instead. URL variants get one canonical URL (choosing a canonical URL); a #fragment needs nothing, because Google generally does not treat fragments as separate URLs.

Limitations

• Search Console never lists rare queries, and the interface exports at most 1,000 rows: conflicts in those queries cannot be seen.
• The checker works from numbers, not from your pages’ text: read both pages before you merge or redirect, and keep what only one of them says.
• A rank tracker’s file has no impressions, so its shares follow the tracker’s traffic estimates, and pages compete by position alone.
• Two pages that answer different needs can rank for one broad query without hurting each other: the recommendation says why, so you can judge.
• Files up to 200 MB and up to 1,000,000 rows a sheet; daily rows beyond 250,000 pairs are combined over the days.

Privacy

On this device: nothing is sent

Your files are read in this browser, in a background thread of this page, and never uploaded: no query, URL or figure leaves your device. Saved projects stay in this browser until you delete them.