There’s a principle in decision theory called MECE – Mutually Exclusive and Collectively Exhaustive. It essentially says that when you define your decision criteria that shouldn’t overlap and they should include everything you are going to base your decision on.
For example, you shouldn’t leave finance out of your decision list because “it will take care of itself.” Finance is part of 99% of all major decisions and so should be included. Similarly, finance should only exist in one part of the criteria and not exist in multiple places. You shouldn’t have Total NPV as a criterion and then the cost of a single piece of software in another. You are double-counting the software in that instance.
This is a difficult principle to wrap our heads around, particularly if you’ve never encountered it before. Putting all of our HR related issues into one box and finance related issues into another seems like we are separating topics that go in hand. You can’t have a project that expands workforce without also increasing costs.
The goal of MECE is to enable us to understand the trade-offs that exist when we make decisions. Expanding workforce may be good but is the decision ranking offset by the incremental increase in costs? Without understanding our independent thoughts on additional capacity versus added costs we can’t score that trade-off.
At the same time, if our criteria overlap and include combined topics we’ll never have a clean evaluation framework for making our decision. The goal of decision theory is to assist in making quantitatively supported decisions. A decision framework that doesn’t provide clean scores will naturally keep us in the qualitative realm.
There are certainly problems with MECE. The first being that it isn’t always possible to separate criteria completely. To use a baseball analogy, if you value players that hit homeruns but also players that drive in and score runs you are double counting since a homerun always leads to both an rbi and run as a result. It is very difficult to separate criteria without going down to valuing player specific attributes such as bat speed and reaction time – the data for which is either not available or difficult to work with.