A ride evaluation platform can evaluate ride services provided by AVs to users, e.g., to determine whether the users are satisfied with the rider services so that the users will request more ride services. The ride evaluation platform may identify and classify operational behaviors of an AV based on a record of the AV's operation for providing a ride service. The ride evaluation platforms can further detect defects in the ride service based on the classifications of operational behaviors of AVs, user expressions that indicate one or more user sentiments towards the ride service, one or more conditions of an environment in which the AV operated for providing the ride service, or some combination thereof. The ride evaluation platform may determine a score for the ride service based on the defects. The score may indicate a likelihood of the user would request another ride service provided by an AV.