Abstract

In this paper we focus on estimating the post-click engagement on native ads by predicting the dwell time on the corresponding landing pages. To infer relationships between feature of the ads and dwell time we resort to the application of survival analysis techniques, which allow us to estimate the distribution of the length of time that the user will spend on the ad. This information is then integrated into the ad-ranking function with the goal of promoting the rank of ads that are likely to be clicked and enjoyed by users (dwell time greater than a given threshold).  The online evaluation over live traffic shows that considering post-click engagement has a consistent positive effect on both CTR, it decreases the number of bounces and it increases the average dwell time, hence leading to a better user post-click experience.