Shadow Backlog

ShadowBacklog

The team was under pressure to deliver a rewrite of the flagship product after having been handed an agressive delivery date. In spite of the Scrum board and other information radiators in their area, the Scrum Master had to deliver progress reports to the CEO every day. The company president and vice presidents attended sprint reviews and grilled the team whenever a story was not closed or sprint goal not met.

One could easily argue that a fixed scope and fixed delivery date is in no way agile – end of discussion – but this still happens in organizations that use (at least partially) agile practices leaving Scrum Masters to navigate choppy waters.

Of course the team developed some coping mechanisms, the most insidious of which was kicking the can down the road.

Whenever the team encountered unexpected complexity or scenarios when working on a story, they said “this wasn’t in the story that we planned, so let’s call the current story done so that our sprint metrics will look good and handle the extra stuff sometime later.” Everyone except for the Scrum Master was happy – for awhile. Team velocities looked good. Stories were being closed as planned. On the surface it looked like features were being completed. The teams were going to deliver on schedule. The Scrum Master knew deep down that they were either going to deliver late or deliver an application with enormous amounts of technical debt. She also knew that the organization could not handle to truth so she soldiered on.

Then the bubble burst. The collective “oh crap” moment happened when people realized that the system should be production-ready but was not. Some features were incomplete. Bugs were still there from parallel testing that had been performed months ago. There was no room in the time before the expected completion date to resolve all of the open issues. Market launch would be delayed.

The ‘shadow backlog’ existed in the developers’ minds and in the list of open defects that were logged elsewhere from ongoing user testing. There were no product backlog items for them and no sprints in the release plan for addressing them.

Reflections:
– Make sure that all remaining work is visible. Create backlog items for stories, technical debt, and bugs. The backlog consists of Product Backlog Items (PBIs), not just user stories.
– Get user feedback during the sprint to avoid surprises later. Ideally users are integrated into the process via acceptance test driven development (ATDD). At a minimum they have an opportunity to see stories in action and provide feedback in the sprint. Any feedback and new ideas that cannot be incorporated during the sprint should be captured as real backlog items and sized/prioritized.
– Don’t be afraid to keep a story open if it was more difficult or took longer than expected when sprint planning was originally done. Better to take a little heat now rather than have the bubble burst later.
– Avoid ‘bucket stories’ near the end of a release that become a dumping ground for stuff that did not get completed in the sprints. Bucket stories can quickly grow exponentially and delay the release.