Machine-learning based multi-step engagement strategy generation and visualization

In a digital medium environment to cause targeted client device users to perform a desired action, a method implemented by at least one computing device, the method comprising:

• receiving a request from an engagement-system user for a multi-step engagement strategy used to control delivery of content associated with a campaign, the request limited to an entry condition and an exit condition for the campaign, the entry condition limited to describing an action performed by a client device user that causes the client device user to be targeted by the campaign, and the exit condition describing an additional action performed by the client device user that causes the client device user to no longer be targeted by the campaign;
• training a machine learning model with data describing interactions of client device users with content delivered in connection with at least one historical campaign;
• generating, using the machine-learning model and without further input from the engagement-system user, a prediction of the multi-step engagement strategy that optimizes performance of the exit condition across the client device users targeted by the campaign responsive to satisfying the entry condition, the multi-step engagement strategy describing a sequence of content deliveries to at least one client device user targeted by the campaign;
• and
• controlling the delivery of content associated with the campaign, in part, by serving the sequence of content deliveries to the at least one client device user targeted by the campaign as described by the multi-step engagement strategy.