Decision-making systems like ActionHub, in their role of optimizing outcomes, are most effective when the choices they make are informed by a clear, singular definition of success, or "optimization metric". Over time, as decision performance is measured, the results can be used to evaluate future choice options for their potential impact on this same optimization metric. Examples of optimization metrics could include product sales, gross revenue, or website engagement. More sophisticated organizations might consider metrics such as CLV (Customer Lifetime Value) or DSI (Down-stream Impact) using predictive modeling and data science to estimate the impact of certain customer behaviors. Regardless of which optimization metric you choose, your marketing strategy can only align to one or risk diluted impact.
ActionHub optimizes for interaction frequency by default, but that is configurable
When generating recommendations, ActionHub assumes that all action types, assets, and labels are equally important, and optimizes for increasing the frequency of these actions and attributes. Therefore, by default, ActionHub recommends actions that are expected to generate the most subsequent actions. For example, a program may be designed to recommend purchases as actions, in which case ActionHub will recommend that a customer make product purchases that are most likely to result in an increase in subsequent purchases.
However, ActionHub is also configurable, allowing for the weighting of certain actions, products, collections, or activities that may be more beneficial to a different optimization metric. Each element of an ActionHub program has an associated weight that measures its capacity to influence the optimization metric. For example, the associated weights for action types, assets, and labels could be based on CLV and a program designed to use purchases, gift registry sign-ups, and marketing opt-ins as actions would recommend the actions with a bias towards increasing CLV.