Most companies consider deploying a pilot system before the full deployment of the entire warehouse. This is not a matter of substituting the first full-fledged data mart as the pilot. The pilot is separate and distinct, with specific purposes. There are several good reasons for deploying a pilot first. The pilot enables your project team to gain broad experience, gain specific experience with the new technologies, and demonstrate proof-of-concept to your users.
If your project team opts for a pilot, concentrate on the basics from the beginning. Clarify the purpose and goal of the pilot and choose a proper subject area for it. Remember that hardly any pilot deployments are throw-away projects. Treat the pilot with all due respect as a regular project.
Embarking on a data warehouse deployment is fraught with potential risks for failure. If you do not strike it right the first time, you may not have a second chance to convince your users about the merits of the new paradigm. Success is a primary goal; you cannot risk a failure. You must be able to demonstrate the potential positive results within a reasonably short time and you must be able to manage this activity quite easily. A pilot deployment to a small group of users in a limited and closed environment is quite appealing.
This does not, however, imply that a pilot deployment is always necessary. Your environment may be different. The group of your users ...