I posted a few weeks ago that I had an opportunity with regards a short piece of work on a Documentum project, its coming to a close now, or should I say the first phase is coming to a close. It has been an interesting experience. The approach taken was:
- perform 4 1-week iterations focussed on specific features
- start each iteration with a demo of Webtop to focus minds on requirements
- end each iteration with a demo of completed work
- allow for time in the following iteration to make changes to the deliverables from the previous
- MOSCOW the requirements as much as possible
Scope was a big problem to manage, as ever, but we agreed up front that we would do no work which required ‘development’, i.e. everything we would do is to be done in DAB, largely Document Types, Lifecycles and Permissions.
What worked well:
- as a way to flush out requirements it was fantastic, we now have a Use Case model which includes requirements which have been delivered but also those which will still require work
- stakeholder buy-in, good buy-in to the solution and the possibilities with Documentum
What did not work well:
- quality of high-level requirements were poor and changed throughout the period. We managed this well despite the challenges but it confirmed my fears that the Object Model was insufficient to progress into this stage.
- lack of environment control. We were not in charge of the environment, it was there when we arrived and for us to deploy into, lack of non-functional requirements were surprising to say the least.
Overall it has been successful and one group within the organisation now have a system they can use as the early adopters while we focus on cross-organisation requirements, both functional and non-functional. Persuading the client to take the step back was hard but necessary and now we have built confidence in our ability to deliver enough for them to trust us. It also provides useful feedback into our process for delivery, a step too far with agile, but a managed success.