Skip to content
DarkCore Grove

Keyword cannibalization across a portfolio of sites

When your own sites compete for the same search, both lose. How cannibalization happens in an affiliate portfolio and how to give each search one article.

Published
3 min read
3 min read
Русский
Українська
Two trees grown into one tangled crown over one patch of ground
On this page
  1. How it happens in a portfolio
  2. Why it costs more across sites than within one
  3. Group searches by results, not by words
  4. Give each cluster one article
  5. How Grove handles it

Keyword cannibalization is when two of your pages try to rank for the same search. Inside one site it splits links and signals between two pages. Across a portfolio of affiliate sites it is worse: your sites compete with each other for the same visitors and the same offer, and a cluster of near-identical pages on several domains starts to look like the doorways Google’s spam policies describe. The fix is a rule, not an audit: each search belongs to one article in the whole portfolio.

How it happens in a portfolio

Nobody plans cannibalization. It comes from working site by site:

  • Each site runs its own research. Two sites about gadgets both find “best robot vacuum for pet hair” and both write it.
  • Writers pick topics from the same keyword list. Without a shared record of what is taken, the same search gets two articles a month apart.
  • Translations and GEO copies. An article copied to a second site with a new intro is the same article.
  • Offers drive topics. When a new offer comes in, every site that carries the vertical writes about it.
A plot divided into separate beds by low fences

Why it costs more across sites than within one

Within one site, Google usually picks one page and the other sits lower. Across sites the effects add up:

  • the visitors split between your own pages instead of between you and a competitor;
  • the work on the second article is wasted;
  • several domains with near-identical pages for similar searches, all pointing at the same offer, read as a network. Google’s doorway policy covers sites or pages made to rank for similar searches that lead the visitor to an intermediate page.
A single key hanging on a hook

Group searches by results, not by words

The first step is to know which searches are really one. Two phrasings belong together when Google shows mostly the same pages for both. Grouping keywords by the search results they share gives clusters that one article can answer. “Robot vacuum pet hair” and “best vacuum for dog hair” may be one cluster. “Robot vacuum for stairs” is another.

A railway turntable seen from above, its tracks leading off in separate directions

Give each cluster one article

Once searches are clusters, make ownership explicit:

  1. One record for the portfolio. A cluster is taken by one article, on one site, for one language and country.
  2. Check before writing. A new topic that hits a taken cluster is either the same article or a different angle with its own searches.
  3. Check before publishing. Compare a finished article with the rest of the portfolio. Similar text means overlap even when the keywords differ.
  4. Merge rather than duplicate. If two articles already compete, keep the stronger and point the other to it.

How Grove handles it

In DarkCore Grove research groups keywords into clusters by the search results they share. A cluster is claimed by one article in the whole workspace, for its language and GEO, and a second article cannot take it. Before publishing, the cannibalization Gate compares the text with the other articles of the portfolio and blocks an article that overlaps. See seven Gates.

All articles