Even given a service agreement, customers and end users must be able to notify the service provider of their needs for specific instances of service delivery. In the CMMI-SVC model, these notifications are called “service requests,” and they can be communicated in every conceivable way, including face-to-face encounters, phone calls, all varieties of written media, and even non-verbal signals (e.g., pressing a button to call a bus to a bus stop).

Regardless of how it is communicated, a service request identifies one or more desired services that the request originator expects to fall within the scope of an existing service agreement. These requests are often generated over time by customers and end users as their needs develop. In this sense, service requests are expected intentional actions that are an essential part of service delivery; they are the primary triggering events that cause service delivery to occur. (Of course, it is possible for the originator of a request to be mistaken about whether or not the request is actually within the scope of the service agreement.)

Sometimes specific service requests can be incorporated directly into service agreements themselves. This incorporation of service requests in the service agreement is often the case for services that are to be performed repeatedly or continuously over time (e.g., a cleaning service with a specific expected cleaning schedule, a network management service that must provide 99.9% network availability for the life of the service agreement.) Even in these situations, ad-hoc service requests can also be generated when needed and the service provider should be prepared to deliver services in response to both types of requests.