ALOHA
JOT Internet Media, one of the leading platforms for acquiring and monetizing web traffic, faces the challenge of optimizing bid value in millions of Google Ads campaigns. This parameter is crucial to ensure the profitability of each campaign.
We implemented a novel approach processing historical data from millions of campaigns with Spark, which allowed us to create a predictive algorithm with scikit-learn and LightGBM. This algorithm not only identifies the bid value that maximizes returns, but it is also designed to be efficiently deployed and monitored using MLFlow and Docker in the cloud (Google Cloud Platform).
The algorithm has revolutionized the JOT strategy, moving from a system based on manual rules to intelligent and automated management. Now, the model accurately predicts the profitability of various bids, selecting the most advantageous one to optimize campaign results.
“By collaborating with WhiteBox, at JOT, we have been able to develop analytical solutions that have allowed us to integrate performance prediction services into the management of our digital marketing campaigns”
INMA
