How Twitter's open-sourced algorithm changes marketing strategy

Executive overview

Twitter released its recommendation algorithm as open source on GitHub. The code reveals concrete engagement weightings that most marketers have never seen before.

Tweet engagement within the first six hours is decisive — replies that spark back-and-forth are worth 150x a like, and relevancy decays 50% every six hours.

Engagement weightings from the source code

• Time on tweet greater than two minutes: 20x a like
• Clicking into your profile from a tweet and liking or replying: 24x a like
• A reply to your tweet: 54x a like
• A reply to your tweet that you respond to: 150x a like
• A report on your tweet: −738x a like

Timing and content decay

• Tweet relevancy drops 50% every six hours
• Early engagement is critical — drive interactions in the first few hours after posting
• Twitter Blue subscribers receive a slight algorithmic boost

Following-to-follower ratio

• The ratio matters to the algorithm — historically gamed by mass-follow/unfollow tactics
• Twitter has since cracked down on automated following behaviour

Implications of open sourcing

• Trust increases when platforms are transparent about how content is ranked
• Copycats and forks are an expected side effect — competitors can build on the code
• Governments and regulators now have direct access to scrutinise the ranking logic