It’s really far between races. I ran a half marathon yesterday, and last time I did it was more than a year ago. It’s not like I’ve passed up on a lot of races, they’re just happening really seldomly around here.

So while I had a baseline measurement - 1:42:26 from last time - it was really difficult to anticipate if I would be able to improve on that time. Basically, anything that I had done to improve from last year has taken 12 months to verify. That’s a long time. And even then, anything that’d affect the performance on that day, e.g. coming back from a cold, a night of bad sleep, and so on, would skew the results.

So while a baseline is fundamental - without it you don’t even know if you’re doing well or not - it’s also important to be able to measure against that baseline often and cheaply. If what you’re doing doesn’t have any impact on the baseline, you can stop that and move on to something else.

This reminds me a bit about how we work with employee satisfaction at Spotify. When I started back in 2013, we used something called Great Place to Work. Every year, the company sent out a massive survey. While it provided a comprehensive baseline, Grate Place to Work didn’t offer us an efficient way to follow up on our improvement actions since it took a whole year to get another set of survey results. Also, it was very sensitive towards anything happening in the weeks prior to the survey, like a failed change initiative. To remedy this, my department turned to a tool called OfficeVibe, which sends a light-weight surveys with random questions every week. This way, we’ll be able to get new measurements every week, enabling us to move faster.

If you want to be a better 5K runner, you shouldn’t just wait for those two or three official races per year, but instead find a route that you can run every other week to see if what you’re doing to improve is working.

Over the past 2.5 months, I’ve done 13 fast runs on my favourite 8.5 kilometer route, and I’m using Strava to graph the trend.


Even though this route won’t give a feedback loop on longer distances like half marathons and marathons, It should be close enough for both 5Ks and 10Ks.