Solving password sharing

Problem/ Opportunity

Currently, Pitchbook uses two criteria to identify a password sharing event i.e., Time difference between two sessions and the change in IP. These are not efficient looking at the increasing number of Pitchbook customers - hence the abundant growth of the user login data and looking at the increasing variety of devices that Pitchbook users use to login to the system.

Solution

We are implementing machine learning(ML) classification models. The first ML model will detect password sharing events in real time. The second ML model will help Pitchbook monitor the accounts for which there is unusually high user usage on the Pitchbook platform.

Benefits & Future Work

Pitchbook will records more meaningful data around the significant features that our ML model identified. The additional data will further improve the efficiency of the ML model in the future; hence it will predict a password sharing event better.