Visualizing all of the user actions that go into the score in terms of their predictive importance can be a very helpful way to breakdown the model and communicate its function with other teams. Plotting each action in a 2D field where Necessity and Sufficiency form the axis gives a view of the relative predictive value of each observable element. I used a dashboarding tool to make an interactive chart that allows filtering by product area, and selecting individual bubbles to see metric details.
Product-Qualified Lead (PQL) Scoring
This post walks through how to build a “product qualifying” model to identify which users of your product are most likely to buy. For explanations of why this is a great idea, and you should already be doing it, see this excellent article from Madkudu, or the Product-Led Growth book.