Has anyone managed to build recursive or self-learning mechanisms into their outreach programs? By this I mean, a reliable way to score leads and messaging so that you best utilize your resources.
I feel like, if you're able to tell an AI model what a good outbound message looks like, based on the ones that historically converted, then it should be able to tell you how close or far away you are from that.
More than just the content of the message, but also all of the trackable datapoints: about the person and their company, their situation, and any relevant context about the lead. Then it can use the successful scenarios to infer how similar or dissimilar the new lead is.
In my head I have some rough ideas of how I'd design a system like this. A lead scoring model that improves over time, and tells you what a good message looks like. Just wondering if anyone else has worked on this angle and how they approached it?