Keyword Cannibalization Tool

Bring paired records

The useful question is not simply whether two pages mention the same keyword. It is whether those pages meet the same need and whether their overlap is costing the site useful traffic or conversions. This report helps you find cases worth investigating. It cannot establish that loss on its own.

Each imported row must connect a query to a page, with clicks, impressions and average position for the same reporting period. Use a Search Console API export or reporting connector that includes both query and page dimensions. Downloading the separate Queries and Pages tabs does not provide those pairs; their totals cannot be joined reliably.

Choose the reporting dates and search type used for your export. Import CSV or TSV, or paste records, then review the column assignments. Keep country and device dimensions when present. Use Search Console’s three-letter country codes and desktop, mobile or tablet labels. Leave a dimension blank for every row if the export does not include it. Do not combine an all-device total with its mobile breakdown.

Read the overlap

A multi-page segment means the same normalized query has impressions for more than one URL within the same country and device selection. Open it to compare the underlying pages. The first page has the most imported impressions, not necessarily the strongest commercial value or the best claim to be your preferred page.

Average position is weighted by each row’s impressions. It is not a single observed rank. The share column divides a page’s impressions by the sum of imported page-level impressions for that query segment. It is not a share of all searches, and it should not be compared directly with a property-level Search Console total.

Use filters to focus on a query, page or minimum amount of evidence. Low-volume overlaps deserve caution: a handful of impressions is a weak basis for a major content decision. Downloads retain the reporting context so another reviewer can understand what was compared.

Look across days

Daily mode adds a date to each query-page record. The report identifies the URL with the most impressions on each observed day, then counts changes between consecutive days with a unique leader. A tie or missing day breaks that comparison. Date-range totals cannot show this history, so leader changes remain unmeasured in that mode.

A changing leader is a reason to inspect the pages, not evidence that Google is confused. Different audiences, changing demand, incomplete exports or two useful answers can produce variation. Read the pages and current results, and compare their broader query coverage before deciding that one should replace the other.

Choose the next step

If the pages serve different purposes, keep those purposes clear. An instructional guide and a product page may both appear for a broad query without duplicating each other’s job. Check titles, introductions and internal links against the actual reader need instead of removing a keyword from every other page.

If two pages are genuinely interchangeable, examine their useful sections, conversions, backlinks and other query traffic before considering consolidation. A page that looks weak for one query may be valuable elsewhere. Do not delete, redirect or noindex it based only on this report. Any migration needs a deliberate destination and a follow-up performance review.

Record the preferred page and the work you decide on in the Keyword Mapping Tool. That separates the observed evidence here from your future content plan. For suspected duplicate URL handling, use the Canonical Tag Checker to inspect declarations, not to assume Google selected them.

Know the gaps

Search Console data is not an exhaustive record of every query. Some queries are omitted for privacy, and the API can return top rows rather than every row. Filters and export limits can hide another page. A single observed URL therefore means only that the supplied records show one page, not that no other page ranks.

Duplicate query-page records at the same date and segment are rejected to avoid counting the same export twice. Aggregate omitted dimensions before importing when needed. Zero-impression rows stay in the full report but do not create overlap. URL schemes, paths, query strings and trailing slashes remain distinct; fragments are removed and reported.

The plugin does not store imports or reports, fetch your pages, infer search intent or change your website. Shared abuse protection records a salted IP hash, tool identifier and usage date. Hosting and security-provider policies apply separately.

Questions

Sources: Ahrefs on diagnosing cannibalization and Google’s Search Analytics API documentation. This tool is independent of those services.