Key Results and the Estimation of Metrics
Nowadays it is almost standard to use OKRs (Objectives and Key Results) in Agile Organizations. It is claimed that this goal-setting-philosophy enables a true delivery/product organization, since the “simple” application of agile working practices like Scrum or Kanban is not enough to align huge working systems towards a common goal. This is funny on two levels. At first, from a general Systems Thinking perspective it is obvious that shared goals are much more effective than focusing on methodologies or frameworks (= processes). A common sense tells us that the systems’ goals (implicit and explicit) are decisive factors if you want to intervene in a system. Like Donella Meadows pointed out in her easy to read paper “Leverage Points – Places to intervene in a system” (1999) mutual targets align a system much stronger than rules or feedback loops. Actually it is the third strongest lever to change a system. According to Donella Meadows only questions regarding mindset have a greater power.
From a cybernetians perspective it is therefore funny that the topic of goal setting and achievement seem to be the hottest thing. Nevertheless I am a big advocate of the OKR philosophy, even though it is heavily hyped. But it is not as easy to adopt. It causes big changes for the people to work with OKRs. This seems to be especially difficult if for decades people were trained not to define their own goals, but instead there was a boss who defined them. One of the strengths of OKRs is the interpretation of Key Results (the numbers) as a range of values and not an exact metric (the problem with this aspects is known as Goodhart’s Law).
Instead of defining a pseudo-precise metric, one thinks in a “from/to range”, which relaxes the mind because it is much easier to think in “min/max values”. It’s brain friendly! Unfortunately, in the “classic” OKR world, people still have to define a number (for the Key Result) – the idea of a range is not visible in its representation. It is a single number. I think this is a problem, so I want to propose a tiny add-on in order to make it easier to apply the philosophy of OKRs.
Micro OKR Hack
The basic idea: Let’s express the metric not as a single value, but use a range of values. This is not only easier to practice, but it also allows us to better cope with the specific complexity of a situation. The width of the range reflects the certainty/perceived complexity by the people who have to estimate a Key Result. At the end of the day OKRs are not a plan, but a bet on the future. From a cybernetic point of view the usage of a range of values increase the variety of the bet (= amount of possible system states). It fits better to the given problem**.
Workshop Key Results: 3 to 5 new ideas
Increase sales by 10% to 15%
Shorten lead time by 2% to 7%
But my OKR tool can handle only single values!
If that is the case, then you calculate the mean value of the range and define the single value above the mean value. Even if one uses the “goal range approach” it still means to be ambitious while setting a target. The basic philosophy of OKRs should be preserved.
Finally: The VSM perspective
If you have been waiting for the second funny aspect about the hype about OKRs, this paragraph is for you. If you look through the glasses of the VSM (Frame + Lenses*) it is obvious, that typical iterative and incremental approaches like Scrum or XP are pretty good on a tactical level (System 1 – 3*). Also the normative part of System 5 is covered in Scrum by having a strong (product) vision. Nevertheless these iterative approaches lack strategic qualities and exactly this is the point where OKRs close the gap. They take care of the System 4 activities and help to align the working system towards mutual goals. The balance of all guidance levels (S3, S4, S5) is essential to enable a viable system that can inspect and adapt as a “whole”. Then a system is in control to steer itself.
* Thanks to Ivo Velitchkov for this metaphor.
** Ashby’s Law of Requisite Variety