Cannibalization skill
The playbook Moonrank's SEO Agent uses to tell a real conflict from two pages that are simply both useful — and to stop the reflex 301 that destroys value.
Download the methodology, see how the agent classifies a conflict, or run it live inside the SEO Agent.
- Category
- SEO
- Agent
- SEO Agent
- Format
- cannibalization.md
- Price
- Free to download
The skill file
What is the Cannibalization skill?
The Cannibalization skill is how Moonrank's SEO Agent decides whether two pages ranking for one query is a problem at all. It reads Search Console by URL, finds the queries surfacing more than one of your pages, classifies each conflict into one of four types, and reaches a verdict — strong candidate, investigate, or harmless — from three facts: how many of your pages compete, the best position any reaches, and whether Google rotates which one it shows. Each conflict is then priced as clicks and value at risk, so the list is ordered by what it costs rather than by how many pages it involves.
The fourth type matters most: the SERP triad, where two pages hold two positions on one query because they answer genuinely different sub-questions. That looks identical to cannibalization in a spreadsheet, and merging them is a straight loss.
It's written as a SKILL.md, a plain-language instruction format that people and AI agents can both follow. Read it to diagnose your own conflicts, hand it to an agency as a brief, or let the SEO Agent run it on your Search Console data.
See it in action
“My main service page started dropping right after I published a related article. Should I delete the article?”
No — this is a same-intent conflict, not an exact-keyword one. The service page holds position 4.2 with 9,200 impressions; the article sits at 8.7 with 4,100 and a 1.5% CTR, so it is diluting rather than competing outright. Repositioning the article on the night-and-weekend angle exclusively, changing its H1, and linking it to the service page recovers the split without losing the article's own 4,100 impressions.
From input to fix list
- 1
You connect Search Console
Ninety days of query data by URL — the evidence any diagnosis needs.
- 2
Moonrank finds the competing pairs
Every query surfacing two or more of your own URLs, grouped into conflicts.
- 3
It classifies each conflict
Exact keyword, same intent, semantic proximity, or a SERP triad that needs no fix.
- 4
It prescribes per type
Merge, redirect, differentiate or leave alone — with at least three implementation steps.
Everything it checks
Conflict detection from real data
Queries that surface more than one of your URLs, not a guess from titles.
Four-way classification
Exact keyword, same intent, semantic proximity, SERP triad — each with its own fix.
Winner identification
The stronger page named on position, impressions, clicks and CTR.
Strategic role check
Pillar versus satellite, and which page is closer to revenue.
SERP triad detection
The false positives — two pages holding two positions on purpose.
Concrete implementation steps
At least three per conflict, down to the H1 rewrite and the anchor text.
When to reach for it
- Explain why a page slipped after you published something similar.
- Decide whether to merge two pages or differentiate them.
- Stop an agency from 301-ing a page that was earning its own traffic.
- Audit a content library before consolidating it.
- Find the queries where you are bidding against yourself.