Complexity is hard to define. The words origins come from Latin ‘complexus’ that signifies something ‘entwined’ or twisted together’; the Oxford Dictionary defines something as “complex” if it is “made of (usually several) closely connected parts”. In common usage we have also added to this definition, including something that we have a lack of understanding of as being complex.
From our experience companies recognise a complexity problem when they realize that they are unable to manage something effectively, usually after recognition that there is a need for change to react to a market dynamic.
Good complexity enables improved performance as your company grows typically by adding product, automation or making processes more efficient.
Bad complexity is created once a tipping point is reached and slows a company down, increasing inefficiency. After passing the tipping point the cost of complexity accelerates.
Complexity is driven by two factors:-
- The number of variations.
- The level of interdependence between the variations.
The impacts of complexity are magnified by:-
- The degree of change required.
- The understanding you have about the cause of the complexity.
There are two perspectives of complexity TOTEM likes to consider.
1) A logical perspective
e.g. I understand my car well enough to get from point A to point B most of the time, but I don’t understand how to fix it because that’s too complicated.
In the example above we usually choose to abstract ourselves from the problem by hiring a mechanic. Luckily for us most cars use very similar core technology and processes so the point at which we can hand over the problem to the mechanic is pretty high level. E.g. There’s a noise in the engine.
For a business complexity is exactly the same. A business becomes overly complex when you don’t understand a problem or how to create change at reasonable cost. When you build your company you often don’t focus on ensuring that the organization, core technology and processes are as simple as possible. By necessity you will have a unique organization performing specific processes and a level of technology customisation. Often the mechanics who built your organisation are no longer available so the understanding of what and how to change is even harder.
2) A mathematical/systems perspective
Sometimes it’s easier to consider complexity from a mathematical perspective and a reasonable level of understanding can be achieved using some simple examples of statistics and probability.
1) Complexity is driven by variation.
- Consider the roll of a dice.. There are 6 possible outcomes.
- If you roll 2 dice there are 6X6=36 possible outcomes as seen below.
- If you roll 3 dice there are 6X6X6=216 possible outcomes.
2) The level of interdependence between the variations.
When we create relationships between different entities we often take shortcuts that binds the two entities together and increases the complexity of the whole system. Instead of having multiple separated systems with dependencies we create one large system with many possible outcomes.
So is there a way we can reduce the complexity?
Yes. By creating a degree of independence i.e. abstraction between appropriate parts of a system.
If we consider the system of 3 dice and instead of rolling all of the 3 dice at the same time we consider them as being independent we can reduce the number of possible outcomes.
i.e. First roll dice 1, and note it’s value. Then roll dice 2 independently, and then dice 3.
This complete 3 part system creates the same number of possible outcomes. However each part is independent and has only has 6 possible outcomes we need to consider. Therefore the total number of outcomes is 6+6+6 i.e. 18.
Please note that this assumes that there is no relationship between to outcome of the first dice roll and the second i.e. there are no rules between them.
An example of this would be a company that has 6 products and accepts payment by 6 methods/channels.
As long as the price for the product is independent of the payment method the two components, product price and payment type can be separated.
If you link product price and payment together you exponentially increase the complexity of the system.
In a system each potential outcome option requires the implementation of data entry, processes etc. and you not only have to set up a system to manage this, but also manage the possible errors which could take place if something goes wrong at any stage.