Extending a service to the whole network immediately can make it difficult to correct design errors. A limited test allows you to observe operation and benefits before taking on broader commitments. However, the pilot must be built to answer precise questions.
I propose to choose participants with characteristics relevant to the goal, declaring the limits of the group. A service tried only on very structured enterprises may not show the difficulties of smaller realities. Results should not be automatically generalized.
Before starting, it is necessary to define what will be observed: time of use, errors, necessary assistance and benefit for the member. The costs incurred by the central structure are also part of the assessment. A free service during the trial may encounter a different usage when a contribution will be required.
At the conclusion, I propose to distinguish results obtained, correctable problems and hypotheses still open. The extension can take place in stages, maintaining a moment of revision. The pilot does not serve to confirm a choice already taken: he must also be able to show that the project must be modified or stopped. This possibility makes the evidence useful for collective decision-making.