Sprints are often introduced in industrial projects with high hopes and little structure. Three months later the team holds daily meetings, the backlog is a long wish list, and nobody can say when the software will be ready for commissioning.
Iterative work needs as much discipline as a schedule, just a different kind.
The cycle of a two-week sprint
- Sprint planning, 2 to 4 hours: the team chooses a sprint goal and the backlog items that serve it.
- Daily meeting, 15 minutes: progress toward the goal, blockers named, no status reports.
- Sprint review, 1 to 2 hours: the working increment is shown to users and to the hardware leads.
- Retrospective, about an hour: one or two improvements for the next sprint.
- Backlog refinement, up to 10 percent of capacity: items prepared for the next sprints.
Ready and done
Two short definitions decide whether sprints deliver.
An item is ready when it has acceptance criteria, is estimated by the team, fits into one sprint and the signals it needs are in the frozen interface list. In industrial software the last condition matters most. An item that depends on an open interface is not ready, however urgent it is.
An item is done when it meets its acceptance criteria, its tests have passed, the code is reviewed and merged, and the documentation is updated. “Done except testing” is not done.
Plan from evidence
The team plans in story points, a relative measure of effort. Speed is the number of points actually done per sprint.
A worked example: the team planned 40 points and finished 35. Over the last three sprints it finished 32, 36 and 35 points, an average of 34.3. The next sprint is planned at about 34 points, not 40. Plans based on hope create a backlog of broken promises.
The burndown shows the remaining points per day. A line that stays flat for several days is a signal for the daily meeting, not for the review.
The product owner says no
Every iterative workstream needs one person who orders the backlog and accepts the results. In a plant project this is often a controls engineer close to production.
The product owner’s most important word is no. New wishes go to the backlog and compete with everything else. Wishes that change the interface list go through the change process of the project.
Connect to the master schedule
The sprint is inside the master schedule, not instead of it. Each integration milestone, such as “software on the simulation model”, becomes the goal of a specific sprint. If the speed shows that the milestone is at risk, the project manager knows it weeks before commissioning starts, while there is still time to act.