So we have a bunch of indicators suggested from our senior leadership team as measures for a program health check. One of these is something we’re calling “Failure Demand”.
Over the last couple of iterations, we’ve got a first glimpse at this – we know it’s one of the trends we wanted to track, so we started with a simple question to each of our teams:
“How much time (as a %) did your team spend on non-feature work this iteration?”
We asked the teams to keep a track on a day-to-day basis and prompted them with a reminder at the end of the iteration. Our first check ended up with a range from 20-60% of people’s time spent on work outside the agreed, valuable feature set they’d been issued e.g.:
- Defects in existing code that needed to be fixed.
- Maintenance work on our platforms.
- Faults diagnosis / 2nd-3rd tier operations support for our platforms.
- Answering ad-hoc requests from people who kept our email addresses over the years and know who to call to get a problem fixed.
- Trading out a team member – in some cases we have SMEs that we still need to loan to another team, but they won’t contribute to their team’s goal, so we can calculate them as a % on non-feature work for that team (imperfect yes, but it highlights a “failure” of us to cross skill, or ensure features are discrete enough that any team can work on them independent of other teams).
- Work in the communities of practice / presentations / brown bag sessions / training courses (not failure demand really, maybe something else, but we don’t know how to track it yet).
Then today I thought maybe we could get a bit more visual than simply displaying the average in a bar graph… and I remembered these box and whisker diagrams from high school. A few more searches and we had a template to build a box plot in Excel.
So here’s our first chart.
We want this graph to converge over time and trend downwards, i.e. trend towards 0% for most teams.
And here’s the box plot template for failure demand if anyone wants it.