User Acceptance Testing Instead Of Manual Testing, Sometimes It Works

After deep analysis two main risks for the project appeared -- normal completion of the whole project itself and elaboration of clear SRS.

The strategy for the key risks was mitigation. We had spent almost 1/3 of the total management cost for elaboration risk strategy and prediction of effects.

Near to the end of schedule when the system was implemented (all 130 requirements were tested) we moved to stress and volume tests. 50 to 200 staff-hours were spent for those tests.

Then "acceptance testing" followed. As mitigation for key risks we spent 150 (!) staff hours for acceptance testing. The whole system was deployed on a test-platform and provided to the customer. As planned, the customer began to make different remarks here and there, which were not in the SRS and were not important for functionality itself, but should be fixed any way as the customer said. With automatic and manual testing we found around 50 bugs, but the customer found around 100 small and big remarks. All his issues were registered as bugs and were fixed ASAP to start testing again.

So, in total, risk management was correct and helped us to develop and send that system.

The 150 staff-hours that were spent for acceptance testing were nothing near the big loses that could have happened to our company if we had done manual and automatic testing.

As a result, the customer received his system and our company spent only that money that we planned at the beginning.