You Manage What You Measure

I've been cycling to work for the past few months. As it gets deeper into Autumn and the temperature dips closer to zero Celsius , making the right clothing choices right is becoming more important.

The hardest part isn't keeping warm: it's making sure you don't overheat. The obvious temptation is to put on as many layers as possible so that you're comfortable when you get on the bike. But then, 10 minutes down the road you start to sweat, the sweat soaks your clothes and combined with the windchill, freezes you.

I've never consistently cycled in a Canberra winter before. So how on earth am I going to work out what I need to wear? Simple: I've started tracking the temperature, my clothing and how comfortable I was for each ride. That way I can check the weather, then look up similar conditions in my dataset and get a good idea of what has worked in the past, and what hasn't.

This idea indirectly came from my last job. The project manager and I were concerned about the amount of work our team was delivering, but didn't have any solid metrics to go by. After quite a bit of thought we came up with a way to measure the percentage of the team's time that was spend directly delivering on the work program. It was quite a hard metric to collect, but it was the one that mattered the most.

That you manage what you meaure cuts two ways: if you want to manage something, you need to measure it. But also you'll start to manage whatever metric it is that your collecting, even unintentinally. We knew this and didn't want to fall into the trap of managing something we could measure, rather than measuring what we wanted to manage.

For example, we could have chosen lines of code or service requests completed. Both would have been easy to calculate and would have given us a rough idea of the state of things. But improving those metrics wouldn't necessarily mean the team were performing any better. All other things being equal, what we really wanted was for the team to spend as much time as possible delivering on the work program, so that's what we measured. Of course that measure is open to abuse too, but overall it provides a good balance.

I wasn't around long enough to see if those metrics helped improved the team's delivery rate, but it certainly focused discussions on where time was being spent and how things could be done differently.

Now, when I don't have enough knowledge about something, I do my best to measure it. For cycling clothing that meant taking 15 minutes to set up a Google Docs form where I can easily enter the data for each ride. To manage my sleep I use Sleep Cycle, which records the amount of sleep I get and provides a running average. I use Runkeeper to track the average speed and monthly total distance for my cycling. Driving around with the trip computer set to show fuel consumption is a great way to get me to drive more efficiently.

I'm not a stats nerd (really!) but having the right metrics helps to motivate me and keep me on track. The trick is making sure that I have the right metrics.

comments powered by Disqus