The figure shows the overall personalization architecture. Requests come in from clients and are processed by the Conductor. Scenarios running in the Conductor call out to data providers, which access data from backend content servers, activity graph service, and property service. Processed responses are returned to the client from the Conductor.