Method: Lookalike Models, Custom Pixels, Sound Reporting
We designed our program to test, evaluate, and optimize based on a large number of inputs. First, we created several lookalike models based on a sample of their most- active list members. Within those models, we layered in other targeting methods, like position on the political spectrum, affinity for environmental or human rights activity, and location. Finally, we developed several custom conversion pixels, based on category of action (women’s rights, voting rights, environmentalism, etc.) to help guide Facebook’s optimization algorithm and reduce costs.
We also developed a reporting system designed to feed back immediate and long- term performance from new names. Every unique targeting and creative element became a lever we can use to make performance optimizations.