What are my goals in life? When I’m stuck in the day to day routine, it’s easy to lose sight of my target. How close am I to hitting my target? Am I progressing in the right direction? How can I tell? More importantly, am I measuring the right thing?
Knowing what to measure is crucial for achieving the desired goal. If we measure the wrong thing, we may achieve what is called in statistics and machine-learning as overfitting. Overfitting is when you start optimising for the thing you’re measuring instead of the end goal.
For example, if you’re measuring your employee using punch cards, they know they need to show up on time and leave on time. They’ll make sure they do that. Does that mean they’ll be more productive? No, it doesn’t.
Imagine wanting to gain weight and only measuring the quantity of food you were consuming. If you’d eat nothing but lettuce (and loads of it!), I guarantee you would not be gaining weight. Clearly measuring food quantity alone would be a poor measurement towards the goal of gaining weight.
When kids at school are measured by the results of their tests, they’ll optimise to passing their tests. Will it help them understand the material better or think for themselves? Not necessarily.
So what can we do? I have two suggestions. First, make sure you’re measuring something that has a causation relationship with your goal, not a correlation. For example, let’s say you’re running a marketing call centre. Measure your employees by the quality of their leads, not how many calls they made. If you measure the number of calls they make, they’ll try and keep their calls short so they could get more calls done. If you measure quality leads, they’ll put an effort towards making successful calls.
My second suggestion is to keep changing what you’re measuring. This is similar to testing your results against an unknown data set in machine-learning. If trying to gain weight, start measuring the number of different types of food you eat. Then, measure the number of meals. For students, we’d be better off testing them with unknown material to which they need to apply their learned knowledge.
Hopefully, with these tools, you can work more effectively towards meeting your goals. And remember: the punch card is a lie.
This is one of the best books I’ve read on measuring things. Once you read it, you will find yourself measuring all sorts of things you never thought about before in ways you never imagined:
https://hubbardresearch.com/publications/how-to-measure-anything-book/
It’s totally surprising how little data you actually need to measure something versus measuring nothing.
LikeLiked by 1 person