Datagran has introduced a new feature named "Facebook as an Action" within its Pipeline tool, enabling businesses to leverage machine learning (ML) models to create custom audiences on Facebook Ads, thereby reducing customer acquisition costs and increasing retention. This feature allows users to export custom audiences derived from various data sources, such as CRMs and cloud warehouses, directly to Facebook, enhancing the efficiency and effectiveness of ad spending. Datagran's CEO, Carlos Mendez, highlights the cost savings, noting a 72% reduction in advertising expenses when using custom audiences. Users can set up these audiences by integrating data from sources like MySQL, Shopify, and Salesforce, and then exporting them to Facebook to create targeted ads. The process includes defining data identifiers, mapping data, and ensuring correct formatting before pushing the audience to Facebook, with the ultimate goal of improving ad targeting and creating value-based lookalike audiences.