Bottom line: Just say no to COTS gap fillers.
Yesterday a small group of us met to identify actions we could take to keep us on track to deliver the capability we signed up for in Release 3 per the Revision 1 baseline.
In December we decided to use a simple burndown chart for Release 3 to show our progress in completing the planned work and work identified after the release planning review that we thought needed to be done to deliver Release 3 capability.
We stole the idea from Mike Cohn. Please take a minute to read his post at http://www.mountaingoatsoftware.com/scrum/alt-releaseburndown. His final image shows how one can use the intersection between lines fit to the tops and bottom of the bars can help estimate when the work will be completed.
Here’s what our burndown chart looked like after two sprints:
The two lines never meet. The capability never gets delivered. That’s not what we want.
The good news is we put a simple tracking tool in place before we started the first sprint. After two sprints, it told us we added more work than we completed. It told us we needed to do something different if we wanted to deliver on our commitment.
What did we do? We identified the causes pushing completion to the right, identified possible actions to pull the completion back to the left, discussed, and decided on concrete next steps.
- deviating from our vision to implement COTS and change business processes as needed to do it
- trying to solve COTS capability gaps. three specific issues accounted for majority of added story points: serial number case sensitivity, unit configuration name change, part number change
- trying to solve later stage problems as soon as we see them
- trying to solve potential user workload issues
- still finding initial set up stories
- systems issues (not a factor now, but may be after March 6)
Possible actions to pull schedule back to left:
- focus on shipping minimal viable COTS product.
- make front-end decision to intentionally *not* do things to fill COTS gaps
- put COTS gap filler user stories at the bottom of the product backlog as possible future work
- for systems issues, we’re already bringing in expertise to have RUP5 installed in Dev 2 prior to next sprint
- considered bringing in additional Oracle CM/MM SMEs, but decided that it would not help
You might remember that last February, when the President’s Budget was released and we first went public with the planned funding from FY 12-17, I told you the CG-LIMS budget would remain small.
We will not be able to afford to customize the COTS tool, or even over-configure it. Limited resources and a focus on serving customers’ real needs will keep us from delivering any more than is absolutely necessary. We will create no more documentation, no more PowerPoint, and no more software than is absolutely necessary to accomplish the mission.
Over the past two sprints, in our eagerness to deliver the best possible application for our users, I think we took more work on our shoulders than we should have. We lost our laser focus on configuring a minimally viable COTS application. We started down a path of over-configuring the COTS software.
Fortunately, we didn’t do most of the work we identified. We identified it and added it to the Release 3 backlog. We made the red bar longer. Now we realize we need to take a second look at that added work. To keep the project alive, we need to accept we are going to deliver bare-bones COTS system initially. That is what the Sponsor has wanted for years. COTS gap fixes must move to the bottom of our list of things to do.
After discussion among the PM, PMd, Sponsor’s Rep, and Product Owner yesterday, here’s what we decided to do:
- 2/27: Review AgileZen and identify “COTS Gap Filler” stories in Release 3 Backlog
- 2/27: move “COTS Gap Filler” stories to bottom of Product Backlog
- 2/27: Update the Release 3 Burndown Chart to reflect changes in remaining work
- 2/27: remind ourselves that we are implementing COTS to deliver a minimal viable product. (This post is the first reminder)
- at every future planning session: ask “Is this a COTS gap filler”
How will we know those actions move the needles in the right direction?
- By 2/28: The updated Release 3 Burndown chart shows lines that converge by the eighth sprint with reasonable cushion for unknowns.
After March 6th we’ll know whether we need to take additional action if RUP5 is not installed in Dev2.
I know that’s a lot to digest. If I’ve left anything out or created confusion or questions, shoot me a note or drop a comment.