This paper introduces TweetCred, a semi-supervised ranking model that assigns real-time credibility scores to tweets, implemented as a browser plug-in. Over three months, TweetCred was used by 1,127 Twitter users, scoring 5.4 million tweets and demonstrating its effectiveness, response time, and usability. During sudden crisis events, spreading spam, rumors, and fake content on Twitter diminishes the credibility of information shared on the platform.