Sector: Online Tracking Service

Opportunity: Clicktale classifies web users according to their web-browsing habits: “engaged user” “deep diver” etc. There are over 2 billion web users, who access 7.9 billion indexed web pages. This is an enormous customer base, almost as much as the populations of India and China combined. Understanding web visitors’ browsing behaviors is a complex task, particularly when incorporating mouse-over activity. Clicktale saw the need for a superior website-visitor tracking service, and developed one that segments customers into well-defined customer groupings based on their browsing behavior.

Approach:Our Data Scientist took historical website browsing data, including stay time, scrolling actions, mouse movements, and clicks, and characterized each webpage visit with these metrics. He then used clustering algorithms such as K-means, Expectation Minimization, Hierarchical Clustering to determine meaningful and manageable numbers of segments.

Results:The models developed led to three to seven segments of users per website to describe all the visitors to that website. Example classes include the “Engaged User”, the “Bounce User”, the “Deep Diver”, and others. The “Engaged User” reads much of the content on the page, scrolls down, or interacts with the page by clicking on a link or mousing over links to read the details.