How Facebook Ranks Search Results Based on Your Friends' Activity

A method for ranking search results by prioritizing links that your social network friends have clicked on previously.

What it covers

This patent describes a search engine that personalizes results by looking at the behavior of people you are connected to in a social network. When you search for something, the system identifies web pages and checks if any of your friends or connections have clicked on those links before. It calculates a 'relevant click' score based on how often your friends clicked those links compared to how often the links were shown to them. The system then ranks the search results so that links favored by your social circle appear higher, and it can even visually highlight those results to draw your attention.

What it doesn't cover

• —Does not cover ranking search results based solely on global popularity or generic page authority.
• —Does not cover search results ranked without using social graph connections or degrees of separation.
• —Does not cover systems that do not track the ratio of clicks to displays for specific links.
• —Does not cover search results that ignore the user's specific privacy or access preferences.

The clever bit

The system uses the click-through rate (clicks divided by displays) specifically filtered by the user's social graph, turning a generic search query into a personalized recommendation engine based on peer behavior.

Why it matters

This patent represents the shift from 'one-size-fits-all' search engines to social-aware discovery. It was a key part of Facebook's efforts to keep users within their ecosystem by making search feel more relevant and trusted, as people are statistically more likely to engage with content their friends have already validated.

Real-world examples

• 1.Facebook Graph Search
• 2.Personalized news feed content ranking
• 3.Social-aware search result highlighting