Embracing the Rollercoaster: The Magic Behind Variation

Is randomness really out of our control?

Variation is our daily driver. Stock markets crash and burn or they go to the moon, your favorite politician is today’s favorite and tomorrow’s most hated. Every aspect of our lives is affected by variation. Is that scary or is it fun?

Just like with everything else, at some point we just accept that’s how things will be. Even though we can’t really explain the reason for every change, it becomes routine. We accept this constant change as long as it makes sense to us. But it gets complicated when we are close to the data.

Businesses change every day, week, month, quarter and year. One day a company could be the industry leader, and next day they throw out a divisive marketing campaign. Customers might be devoted fans, and then become determined boycotters. This is normal.

When we accept change as a constant, we don’t really need to explain every single up and down. We know the seesaw goes up and down. It would be strange if it went sideways. Only then should we actually be interested in finding out why. But usually businesses still care about explaining the all too familiar change.

Why did sales plummet? Why did sales grow? What did we do during that week? OK do more of that and less of this. Variation is scary because it brings uncertainty, so we try to feel in control however we can (even though there are things we can’t change).

There are two types of variation. One comes from the way data was collected (measurement variation), and the other comes from the randomness of the process itself (random variation).

This is more important than you think. I know it sounds kind of boring and useless, but this opens the question “Are the decisions being made based on random variation, or is the variation based on an underlying process?”. This is suuuuper important because it determines if the variation we’re dealing with is under our control or not. If it’s just random, well then what can we do, but if it’s not, we definitely can control the measuring processes.

For obvious reasons we’ll always want the latter. But if it’s not so obvious why, imagine going to a casino. You play blackjack and feel hopeless when you lose money because “Well I can’t really control luck, can I?”. But then you realize that the variation can be controlled by counting cards. Now imagine those savings/earning, but for a company (millions of dollars instead of thousands).

The only constant, is change