Researchers from Pinterest and other institutions have developed a new method for improving the performance of ads ranking models. They constructed a large-scale graph that combines users' onsite interactions with their offsite conversions. A new model called TransRA was created to integrate this information into ads ranking models, leading to significant improvements in click-through rate and conversion rate predictions. The framework has been deployed on Pinterest's Ads Engagement Model and resulted in a 2.69% lift in click-through rate and a 1.34% reduction in cost per click. --- Why it matters: This work matters because it shows how large-scale industrial models can be improved by incorporating offsite conversion data into ads ranking models, leading to better performance metrics such as click-through rate and conversion rate. Source: https://arxiv.org/abs/2508.02609