A pilot is often used to test the design of the full-scale experiment. The design can then be adjusted in time. This can turn out to be valuable: should anything be missing in the pilot, it can be added to the experiment and chances are, that the full-scale (and more expensive) experiment will not have to be re-done.
Often in Engineering applications, pilot experiments are used to sell the product, and provide quantitative proof that the system has potential to succeed on the larger (full) scale. Pilot experiments are also used to reduce cost, as they are much less expensive than the full scale system. If there is not enough reason to provide full scale applications, pilots can generally provide this proof.
In sociology pilot study can refer to a small scale research before the main research is done. This will prevent errors from happenning in the main research.