100 Million Users and One Missing Segment

This is the history about a combination of machine learning, census data, and a healthy distrust of offline metrics helped Globo.com solve one of its biggest advertising problems. Back in 2018, Globo.com had what many companies would consider a nice problem to have: we had about 100 million users. Unfortunately, advertisers weren’t particularly interested in buying impressions for “100 million humans of unknown characteristics.” They wanted targeting. And the most basic targeting segment of all, the gender, was exactly where we had a problem. The Problem: 100 Million Users, 5 Million Labels At the time, around 10 million users browsed Globo.com while logged in. Sounds great, right? Well, only about half of them had filled out their gender in their profile. To solve the problem, the Profile & Segmentation team trained a machine learning model using the browsing behavior of those 5 million labeled users and applied it to everyone else. The model looked promising offline. Which, as every…