How Do We Improve Our Improvement Process?
One thing we talk about with agile is the idea of continuous or relentless improvement. Most organizations want to improve, but like a lot of things it is sometimes hard to feel good about the improvements you are making. Further it is hard to establish a continuous practice where the whole organization is doing these improvements and feeling good about it.
One tool that I've seen helps organizations is to treat improvements as experiments and talk about them this way. The traditional approach to capturing improvement is to set up SMART goals. The problem I have with this approach is that we are treating the improvement as a conclusion (it will result in the improvement we expect - its deterministic) rather than recognizing that the new approach might in fact result in no improvement. And when this happens the thinking is that you have “failed” because you did not meet your goal.
Learning is not failure. And to capture this instead of talking about a change or an improvement we set up an experiment.
To define a good experiment we need a definition of what the experiment is, what the expected result is, what the result actually turned out to be, what we learned, and what we will do as a result. This is the scientific method.
And I would then maintain a list of experiments we have run somewhere in order allow others to learn from our experience.
So, for example, during a transformation to agile, we might want to understand whether the idea of a “True Team” (see Why Do We Form Teams When We Transition To Agile?) actually is a concept that works in our organization. We could setup and experiment as follows:
- Experiment: Does a “stable team” result in increased engagement (with the work and with other team members) when team members are in multiple locations but in compatible time-zones?
- Label / category: Stable Team
- Expectation: Team members will report improved engagement from team over their traditional working method as a result of moving to a stable team structure. Concern is that “remoteness” will cause a reduction in team engagement.
- Metric: Survey questions (note: example provided for discussion only – expect will need some better design):
- Question: “In comparison to how you have worked in the past, how would you describe you level of engagement with other people working on the team?” – “We are not working well together”, “We don’t seem to be working together as well as we have in the past”, “I’ve seen no change in how we work together”, “We seem to be working a little better together than we did in the past”, “We seem to be working together much better than we have in the past”. Plus a comment field “If possible, please provide a specific example that supports your view.”
- Question: “In comparison to how you have worked in the past, how would you describe your understanding of the goal of the work?” … or perhaps this is a separate experiment …
- Timing: Do survey 4 (say – need to set up in time for rollout) weeks after team kick off, analyze results and provide feedback one week after that
- What Did We Learn?: TBD
Then, when the time box is finished, we collect the result, and determine what we learned and what we do next. (Hopefully in this case you got the expected, positive result:-))
Experiments can be done at all levels. Teams can set them up as part of a Retrospective. Transformation coalitions can set them up as a result of their Retrospectives. If you are doing SAFe experiments would be generated as part of a the Inspect and Adapt. And so on.
Using this approach at whatever level you operate at helps you become a “learning organization” where experiments are seen as a normal part of the work you do. The vocabulary will help and assists by making change less threatening (more “try it and see” and less “you will do it this way”). In particular we want to treat new understanding as a “success” (and publish it as a “win”) to aid the momentum. It also like to encourages a spirit of experimentation beyond just “lets try it!” (said in giggly tone). Experiments should be aimed at generating useful knowledge.