Same-day delivery of online orders is becoming an indispensable service for large retailers. We explore an environment in which in-store customers supplement company drivers and can take on the task of delivering online orders on their way home. Because online orders as well as in-store customers willing to make deliveries arrive throughout the day, it is a highly dynamic and stochastic environment. We develop two rolling horizon dispatching approaches: a myopic one that considers only the state of the system when making decisions, and one that also incorporates probabilistic information about future online order and in-store customer arrivals. The results of our computational study provide insights into the benefits for same-day delivery of this form of crowdshipping, and demonstrate the value of incorporating and exploiting probabilistic information about the future.