Ad Quality Analysis : Data Scientist Internship Report

In this report, I detail my internship work at Criteo as a data scientist, focusing primarily on optimization strategies rather than developing new AI methods. My main contributions centered around the Ad Quality domain, involving two key projects: Ad Quality Score (AQS) and Ad Fatigue. Regarding the AQS initiative, the causes behind the low AQS are unclear, prompting a study of score patterns to uncover them. Additionally, we encountered problems with the accuracy of Landing Rate (LR) data, which we were committed to improving. Moreover, the closely matched proportions of various closing reasons posed a challenge in identifying the main factor responsible for banner closing. To address this issue, I introduced two methods aimed at providing more precise closing reason assignments. Regarding Ad Fatigue, uncertainty exists about its presence among Criteo users and the appropriate measurement approach. I conducted an analysis of frequency patterns among users who encountered multiple ads, aiming to identify potential indications of ad fatigue.