Handling Support in Agile Project Management

Mike Cottmeyer from VersionOne has written an interesting blog post about the problem of handling support in agile software development teams. Mike highlights three potential approaches for handling this problem. Personally I think there is a fourth approach, as follows…

  • Make sure one person (probably a manager of some description) is responsible for triage – that is, deciding whether or not a support issue is really priority enough to interrupt the current sprint, or whether it should be deferred and scheduled in the next or a future sprint.

  • If it does have to be handled in this sprint, respond accordingly without estimating the work and without trying to break it into tasks.
  • Allow it to impact your team’s velocity as naturally it will.
  • When you do your next round of Sprint Planning, plan based on the velocity you achieve, so an allowance for support is naturally incorporated into your level of commitment. Do not allocate any specific time for support.
  • If your velocity is unstable and fluctuates significantly, try calculating standard deviation on your average velocity. Standard deviation will give you an objective way to understand the level of risk you are taking if you plan on your average.
  • If your standard deviation is high, the risk of not delivering is high. In this case, commit to a lower velocity and prepare some stretch tasks to be included if you have time.

Of course if you are running a project across multiple sprints, this will still affect your release plan. But it should give you better predictability, and allow you to take action or manage expectations accordingly.

Kelly.

5 Responses to “Handling Support in Agile Project Management”

  1. Dennis Stevens says:

    I like this approach. A product of uncertain quality is likely to have a relatively consistent flow of defects back to the development team. It allows for predictability of performance. And when the team does produce lower defect code it exceeds expectations rather than falls short. This is good from a Project Management approach as well as a team learning approach.

  2. Anonymous says:

    In our team we have a dedicated pair time for 6 of the 8 working hours (i.e between 9.00-16.00) leaving room for support cases, code reviews, meetings regarding other projects etc etc in the non-pair time.
    The pair can of course choose to continue to work on their task during the non-pair time if no other bookings or tasks of the support nature has showed up.

    The project manager together with the support staff decides on the hairier cases whether a pair needs to be interrupted in order to address the support issue immediately.

    It gives the day a rythm and seems to be working.

    This means of course that all of our estimations are done on the basis of a pair day being 6 hours long.

  3. Ed Darnell says:

    Is there a more agile solution?

    For me the maintenance issue is more a symptom of waterfall thinking, than true agility. Defects are some of the best feedback you can get. Agile teams thrive on feedback.

    A defect is an indicator that there is something not quite right. The knee-jerk (non-agile) reaction is to look to process or roles to solve the problem.

    The agile answer is to look at what non-agile behaviours might be causing the problem in the first place.

    Agile teams should be able to take defects in their stride. Test driven discipline should help avoid defects in the first place, but where they do slip through agility should allow them to be rectified quickly and safely.

  4. Anonymous says:

    Ed,

    You are correct that defects (especially in any quantity) are likely symptomatic of another problem. However, that knowledge does not help if your team is responsible for maintaining the collective work product of dozens of programmers working over 20+ years. It would take many years for a small team to rewrite the code in an Agile manner like TDD, so some other solution must be found…

  5. Ed Darnell says:

    Agreed. Legacy code is an interesting issue, and one CIOs worry about with respect to agile.

    It strikes me this is a problem agile teams can tackle. It is a perfect candidate for 80/20 thinking. Most of the legacy code base can be ignored (if it ain’t broke don’t fix it). Defects are clues as to where fragments of re-engineering effort are worth investing.

Leave a Reply

What is 8 + 8 ?
Please leave these two fields as-is:
Please do this simple sum so I know you are human:)

There are 101 ways to approach anything.
To find the best way, sometimes you need expert help

What People Say

“Kelly is an Agile heavy-weight. He came in to assess my multi-million $ Agile development program which wasn’t delivering the right throughput. He interviewed most of the team and made some key recommendations that, when implemented, showed immediate results. I couldn’t ask for more than that except he’s a really nice guy as well.”

DAN PULHAM, DIGITAL DIRECTOR
TELSTRA