Like many other Agile practitioners, I have seen too many cases where an organization wants to adopt Agile, but believes that all they need is a little training. Of course, the most extreme would be when a group believes they can send one person off to become a Certified ScrumMaster and then they can simply train everyone else. While this intuitively sounds foolish, and many people could begin to articulate the shortcomings of this mental model, I’ve struggled to present a clear and succinct view of what exactly why this model doesn’t work. Although I recently came across a very good model that captures what I tacitly knew already. I hope this is valuable to the rest of you out there trying to make the case for coaching.

Thomas Crane, in his book “The Heart of Coaching” identified a model from an article in the November 1979 issue of the “Training and Development Journal”. Sadly, I could not find the original article, but I’ve adapted the diagram to communicate the model. If we think about change on two levels, we see that there are changes in behavior and ultimately changes in results. These two dimensions remain roughly proportional in a static system, as your behaviors impact your results. Things get really interesting once you try to change behaviors. We frequently see this done by sending people or a team to training. Indeed, the “Certified ScrumMaster” classes, are often this point of entry. Now we see that behavior changes, but once someone goes to apply this new behavior, the results are actually worse than when they were using the old behaviors. Many people are familiar with the idea of the “j-curve”, as we use a new practice for the first time, we’re actually quite clumsy and our outcomes are not as good as if we were using an old technique. For those of you who ski, when you first learn to parallel ski, it’s actually much harder to get down the mountain. In fact, when you ski an expert trail, you probably find yourself reverting back to the snow plow. It may be a less refined technique, but you’re effective with it. This is exactly what we see when people first learn and apply Agile practices.

Training with No Coaching

What we observe is that while one’s behavior change at first, they are immediately put back in their prior environment and begin to revert. Perhaps not everyone went to training or the boss doesn’t really care about Agile. As Weinberg observed, the cucumber eventually gets pickled. What’s really interesting as that at the same time, our poor change champion is not performing at a lower level as they try to use new practices. Thus, the “new” stuff actually feels worse than the “old” stuff they were trying to fix. We know some of this is a learning curve, but the observed behavior is that as one reverts back to the prior standard, results are actually getting better. Thus, one can settle at a level with slightly better behaviors and slightly better results, feeling like this is the proper place to be. David Douglas and Robin Dymand talk about a situation they call “we suck less” and I think this nicely summarizes how people can fall into that trap.

Training with Coaching

So how do we help people maintain, or even improve, their behaviors after a training class? Well it must be reinforced, and this is where coaching comes in. Having committed time and energy to learning a new way of doing things, people need ongoing support as they begin to apply them. No matter how good your training class, it will never be a substitute for the messy reality of the world in which we must all operate. Also, we need to remember that there are very real pressures that have driven us to our prior behaviors. We need to begin to apply counter pressures so that we don’t revert back to old habits. This second diagram shows the two dimensions, but now the team receives coaching after they complete training. The coaching helps reinforce practices so that they don’t abandon their behaviors as they move through the adoption curve. As results get demonstrably better, this can create a virtuous cycle where coaching can eventually ramp down and the team becomes self-sufficient, continuing to improve it’s own performance without any external support.

Like many things, this probably introduces more questions than it answers. How long is the right amount of time for coaching? What if the coach is hands on, or the team is supplemented with other experts to get through the adoption j-curve? Are there other ways to change the environment such that teams feel positive pressure to maintain behavior besides through a coach? These all probably merit further discussion, but let me leave it here for now. So I am curious, what are you experiences with rolling out Agile, or any type of change, initiatives? Is this model consistent with your experience?