Predictable Uncertainty

Uncert

See also

• http://itmanagement.consulting/
• 3 Common Questions Scrum Masters Are Asked

ainty

In the current economic world were many organizations are operating on a quarterly basis the idea of predictable delivery for any 3-month period is highly desirable. Let’s look at an example where we have a level of 30% uncertainty averaged over a 6-week period. What this means is that 30% of all the active work items currently within the 3-month window of upcoming work was created within the 3-month window when it was added to the roadmap. This is something that is less desirable in a deterministic approach to the quarterly driven economies. With my pragmatic eyes though, and in my experience, a level of uncertainty on department level of 30% for a 3-month period is not bad. Do note, however, that this does NOT say that items are actually completed, just that the plan is to start.

Looking at the trend for the average uncertainty per week in our example we see a somewhat upward trend, with a peak reaching as much as 38% in week 1947.

The point I want to make, is that should make sure to start planning with this uncertainty in mind and not be surprised when you can’t deliver on 100% commitment months in advance. You can still aim for that, but you won’t get there by ignoring the available data.

Speaking of data, where could you get your hands on this kind of data? This is simply extracted  from the roadmap by looking at it, given that you have fields for created date and end date. If your aim is to have a roadmap that is your single source of truth, this must be the best data there is.

Some of the changes that are depicted in these numbers will be easily avoidable, and some aren’t even changes to the original plan, e.g. when breaking down a large work item into smaller work items. The question to ask yourself before trying to use those arguments, however, is; will you be more, or less predictable by incorporating this number in your planning as is? The 30% that is.

Predictable Uncertainty

The challenge of working with complex problems is that you don’t know what you don’t know. This is the uncertainty I am talking about. The difficulty of planning in an environment where you are dealing with complex problems lies in this fact, that you don’t have all the facts, and that you can’t have all the facts before you act. The operating mode in a complex environment is Probe-Sense-Respond.

Probe means action, that you do something, preferably define an experiment where you have thought about what you believe will happen as a result of the action.

Sensing means you look at the actual results to understand what impact you had and if it was the impact you believed you would have, if you achieved the desired outcome.

Responding to this impact is the last step and likely to be an adjustment to the plan, tweaking your expectations or going a completely different direction and designing another probe.

In retrospect you can also look at the historical data to see how good you are at predicting the impact you will have with your actions in an attempt at designing “better” probes. What we are looking at above is an abstraction of the results from these actions in retrospect. Meaning, the number saying 30% above, is the current way of predicting the level of uncertainty you have, i.e. how much you need to adjust the plan.

What does this mean in terms of the road metaphor?

If we go back to a simple version of the road where we assume that we can have 100% utilization while travelling at the allowed speed limit, effectively ignoring the nature of complex problems, we will have the following situation.

If we just for a second accept this deterministic view of planning, we’ll try to clarify where the 30% comes into play. This picture would be true if we had 0% uncertainty in our plan, but we don’t, we have at least 30% in pour example above. Here is what that looks like.

The question to ask here is, where did those vehicles come from? The criteria I used when gathering the data was “active work items currently within the 3-month window of upcoming work that was created within the 3-month window when it was added to the roadmap”. That’s not what we are looking at here. That means that the vehicles has alternative ways to end up on this stretch of road. There must be connecting roads along the way.

This would be a more accurate depiction of the state of our road. The green arrows show where we currently have oversight and understanding of the traffic merging onto the road, the red arrows represent where smaller roads merge onto the road where we lack oversight. The green arrows account for 70% of traffic and the red arrows account for a total of 30% of traffic. With this example we are also saying that we don’t know where the smaller roads merge onto the larger road, since we lack oversight. We are, however, drawing the conclusion that there must be smaller roads contributing to the traffic situation based on the fact that we are measuring a level of 30% uncertainty. That is, when we take a snapshot of the road, 30% of the vehicles will be red as in the picture above.

To begin to combat this issue we don’t have to know where the vehicles are coming from, we just need to acknowledge that they are in fact red. Now that we know that we have 30% red vehicles on the road we can adjust the number of vehicles that can merge onto the road from the green arrows, where we have oversight. We simply decrease this amount to not exceed 70% of the total capacity, leaving gaps for the vehicles that will merge onto the road further down.

Once again, we don’t have to know where nor when the red vehicles will show up, we just need to make sure that we can accommodate them. In fact, given the nature of complex environments, we can’t know when they will show up before they have done so.