Detect keyword cannibalization across your site

The scenario

You have a content site with 200+ blog posts published over five years. You suspect overlap (the writer in 2022 covered "customer interview templates" and someone in 2024 covered "user research interview guides" - probably the same intent). You want a definitive list of cannibalized queries before deciding which pages to keep.

The prompt

Use the gsc-pap MCP server. For https://example.com, find queries with cannibalization in the last 90 days: queries where 2+ URLs each get at least 50 impressions and rank within positions 5-30. For each cannibalized query, list both URLs, their positions and clicks, and recommend whether to merge, redirect, or differentiate.

What happens

1. 1

   Claude calls find_cannibalization

   The tool aggregates 90-day search analytics by query, then groups URLs ranking for the same query. The output is a list of (query, url_a, url_b, position_a, position_b, clicks_a, clicks_b, impressions). Claude orders results by total wasted impressions (impressions where neither URL captured the click).

2. 2

   Claude classifies each pair

   Three patterns emerge: (1) Near-duplicate intent - two posts answering the same question slightly differently. Recommendation: merge into a single canonical URL, 301 redirect the loser. (2) Different intent, overlapping query - one post is broader, the other deeper. Recommendation: differentiate H1s and link them as related content. (3) Old content + new content - the older post still ranks but is stale. Recommendation: redirect old to new and let Google consolidate signals.

3. 3

   Claude suggests the minimum action

   Output sample: "Query: 'customer interview templates' - 4,200 impressions/90d. /blog/customer-interview-templates (pos 12, 45 clicks) vs. /blog/user-research-interview-guides (pos 18, 22 clicks). Pattern: near-duplicate intent. Action: 301 /user-research-interview-guides to /customer-interview-templates. Estimated unification gain: 65-80% click recovery within 4 weeks."

Outcome

Typical mid-size site surfaces 8-20 cannibalization pairs. Most are mergeable in under 30 minutes each (write a 301 redirect, confirm canonical, update internal links). Re-running the audit a month later shows the merged URL climbing 3-7 positions and combined click count jumping 40-90%.

Where to take it next

• ·Run the same query for branded terms ("only queries containing 'example' or your brand name") - branded cannibalization is the easiest to fix and usually has the biggest CTR uplift.
• ·Use historical_position_trend afterward to confirm the merged URL is climbing.
• ·Re-audit every 6 months - new content always introduces fresh overlap with old posts you have forgotten.