Implementation Testing

Implementation Testing

When an automated process is first deployed, it should be rolled-out on a limited number of resources and with a reduced workload, only increasing if performance is at a level anticipated by the Business.

Initially, a pre-agreed and representative proportion of cases should be audited to ensure the process is behaving as expected. As the workload increases, unseen or unexpected scenarios may occur, and this may cause a higher than anticipated exception rate. System exceptions with a high occurrence should be fixed in the development environment and promoted again through the higher environments.

Where realistic test applications have been unavailable in the development and test phases, a longer implementation test period will be needed to ensure the process is working as expected in production.

Blue Prism have devised a successful approach that involves proving the process against live data in an attended manner prior to it being implemented into a production environment. Similar to verification testing mentioned above, this essentially involves an SME observing and manually checking the applications while the developer carefully steps through the process. Any unexpected application responses or missing steps can be immediately configured by the developer and verified by the SME.

This technique can be likened to teaching a human trainee before letting them work unaided, and a process that has been exposed to live data before being deployed into production will invariably require far less live-proving than a process that hasn't. This is due to the ‘fix on fail’ approach and the reduced reliance on manufactured test data in a conventional testing approach.