A voice-controlled device monitors content delivered to a user through a non-interactive or unidirectional medium, such as broadcast television or radio. Each time the device recognizes delivered content that includes an advertisement or solicitation, the device records information identifying the content and the time at which the content is delivered. The device also listens for environmental audio cues suggesting that a user has performed a desired action in response to viewing or hearing the recognized content. If the device determines that a user response has occurred within a predetermined duration of time after a corresponding advertisement or solicitation, the device correlates the response with the content. The device then performs a corresponding activity, such as notifying the content-delivery service that a royalty should be collected from the content's source; directly transferring a royalty from the source's account to the service's account; or updating response-rate statistics associated with the content.