Systems, methods, and computer-readable media are disclosed for optimizing a value of a dispensing parameter of a product dispenser based at least in part on product usage data. The dispensing parameter may include any suitable adjustable parameter of the product dispenser including, but not limited to, a dispense duration, a volume of product (e.g., a shot size), a sheet length, a capability of a user sensor to distinguish between genuine and false triggers of the sensor, a delay setting that specifies a delay time between successive dispenses, or the like. The value of the dispensing parameter may be optimized to align an actual performance outcome with a target performance outcome. The target performance outcome may include, for example, minimizing product usage per user, maximizing user sensor reliability (e.g., a percentage of dispenses that occur in response to genuine user-initiated triggering events), and so forth.