Reexamined: Is that Sprint Burndown really worth it?
I wrote an initial blog post on this topic back on November 19. To my surprise, someone not only read my post, but even spent the time to write a related response of their own. Read Chris Murman's blog post on Scoreboards. Now I suppose it would be cooler if I told you that Chris and I were in violent disagreement, but unfortunately for you carnage-seekers, we're not that far apart. So before I begin - a big thanks to Chris for reading and responding to my post. I appreciate his viewpoint and it's always good to know that someone actually reads my blog posts. His response prompted me to do two things:
I want to elaborate on my original post and ensure I made my point accurately. Too many math & science classes and not enough literature.
Secondly, I will tie it directly to Chris's post and explore some of the other scoreboards available to Scrum teams.
By-the-book Scrum calls for Sprint planning to accomplish the following:
"The What" - The PO & Team review candidate stories for the sprint, discuss specifics, and create their abstract estimate using story points.
"The How" - The Team breaks down your stories into tasks, estimating those tasks in hours, confirming that the total backlog can be done and should be committed to by the team.
"The Commitment" - After the Team has finished task estimation, they can confirm that their initial estimate was correct, or re-estimate stories whose complexity/effort increased significantly during tasking. Once that has been done, they can firmly commit to the Sprint backlog.
The point of my initial blog post is that the time involved in breaking down those tasks and creating those task estimates in hours in by-the-book Sprint planning may not be worth the benefit. The implied benefit being obtaining possibly better estimates and the ability to track a traditional Sprint burndown chart.
Instead, I indicated that some of my more mature teams do not create hourly task estimates and allow "The How" to happen during the Sprint in a just-in-time manner. By shifting “the How” out, you save some Sprint planning time but cannot have a traditional Sprint Burndown chart (as you don't have the hourly task estimates). Because the teams are mature, they still estimate accurately even without the task estimates. However, they have to rely on other scoreboards to measure in-sprint progress.
Generalized Discussion on Scoreboards
The net result of my first post is that those mature teams who opt away from hourly task estimates need to fill the progress void vacated by the Sprint Burndown chart with some other scoreboard. Let’s start by taking a look at the cost-to-benefit of the by-the-book Scrum process including the Sprint Burndown chart.
By-the-Book Scrum Scoreboard
By-the-book Scrum Scoreboard provides the following:
Potentially more accurate estimate as a result of the tasking, time estimates, and doing the math - though it still relies on overzealous, hourly guestimates at its core.
More accurate tracking during the sprint by seeing each task consumed - though this too is only as good as those same hourly estimates.
Longer Sprint planning (less development time/business features)
Modified Approach Scoreboard
Now let's examine the Modified Approach Scoreboard that I mentioned for mature teams:
Shorter Sprint planning (more development time/business features)
Potentially less accurate estimates - though mature teams get really good at "spotting an 8 versus a 5"
Less accurate tracking during the sprint
Given that the most significant cost of the “mature team approach” is less accurate tracking, let’s move on to a more generalized discussion about scoreboards – those information radiators which tell you how you are doing within a sprint.
Below are a handful of scoreboards which I have seen used on Scrum teams. Two of these scoreboards are part of classical Scrum (Sprint Burndown & Task board). The others are scoreboards which I have seen used by teams. I am including here for completeness – not because I recommend them. I am pragmatic though, and see benefit in modifying certain aspects of Scrum provided the spirit of Scrum is maintained.
Cost: Requires the How discussion, tasks for all user stories, and task estimates in hours.
Accuracy: Most accurate depiction of progress during the sprint. However, it is only as accurate as the team's ability to estimate initially, and we know the history there.
Tasks Burndown Chart: A burndown of all tasks (all tasks have the same weight) to produce each user story in the sprint backlog.
Cost: Requires the How discussion, tasks but no estimates are required (all tasks are considered equal).
Accuracy: Given that there is variability in tasks (generally between 1-8 hours) the accuracy of the Tasks Burndown chart is less accurate than the Sprint burndown. However, on average still may provide some indication of Sprint progress.
Cost: Requires only a committed sprint backlog with story point estimations.
Accuracy: This scoreboard significantly lags the effort of the team as many hours are spent before the first story is completed. Additionally, the burndown moves in large chunks each time a user story is completed, further reducing its accuracy.
Task Board (The Scrum standard): A compilation of post-it notes (or the like) for each sprint task grouped in rows for each user story in the backlog – which move along various states of completeness.
Cost: Requires the How discussion, tasks for all user stories and task estimates in hours.
Accuracy: An accurate depiction of progress during the sprint. It does not visually depict progress-to-goal as well as the Sprint Burndown, but you can readily see impediments or a task taking longer than planned.
Story Board: A compilation of post-it notes (or the like) for each sprint user story in the backlog – which move along various states of completeness.
Cost: Requires the How discussion.
Accuracy: A slightly less accurate depiction of progress during the sprint than the Task Board. Like the Task board, it does not visually depict progress-to-goal as well as the Sprint Burndown. Additionally, it does not have the task detail present in the Task board.
In summary, transparency is one of the most powerful aspects of Scrum. The Sprint Burndown and Task boards are very effective at both internal and external team communication transparency. However, the cost to produce these items should always be considered & minimized. The primary goal of Scrum teams is to produce working software – so we should always strive to maximize development time. In some cases, other artifacts/methods can be used to balance preserving the communication transparency with team throughput.