Using Time-to-Click to Identify Websites with Robot Traffic

Abstract

This paper shows how to use statistics to identify automated browsing leveraging time between events, such as clicks on a website. The methodology was applied on web advertising traffic, aiming to identify websites with a high share of non-human traffic by using the time between the timestamp an ad is displayed and the timestamp it is clicked.

Keywords: ad tech, statistics, data mining, fraud detection, bot detection, gmm

JEL Classification: L86