How to understand the ranking of Confluence's Search Results
We've made changes to the Confluence Search ranking logic in Confluence 8.0 and later. Learn about these updates at Ranking of Search Results.
Purpose
When searching for items in Confluence, it may not provide the results in the order you were expecting. For example, you may expect a certain page to be ranked higher than it is in search results; or a blog post does not feature on the first page of results.
The goal of this document is not to explain the search algorithm (as this changes between releases) but rather to help explain the overarching principles in Confluence Search Results.
The Confluence "Explain" Action
Visiting http://<CONFLUENCE_URL>/explain.action?queryString=keywords will provide a formatted list of each search result, and what factors contribute to its overall score. This is a very technical breakdown of how the search results are generated. We've got some rough guidelines below.
Notes regarding CQL
- The Confluence "Explain" Action does not support CQL at this time.
- CQL generates rankings differently to the primary search mechanism, although CQL will be the primary search method in Confluence 5.9 and above on Server and Data Center releases.
What affects a document's score?
All documents have the same chance to appear in search results. Weightings are made up of the following:
Document Age
Newer Documents will be given a slightly higher score than older documents. Keep in mind that doesn't mean that newer documents will appear first, just that they give a slightly higher weight.
Title
Confluence will look at the number of words within a title (excluding minor words, such as "the", "and" etc), and will check how many times a word appears in the title.
Content
As with the title, Confluence will look at the number of words within the content, and check how many times the number of words appears in the content.