Roland Ewald

Software Engineer · Researcher



Fitness tracker accuracy

2025-03-16


Here is an interesting new study on the accuracy of fitness trackers (via HN), putting the overall accuracy across all evaluated devices at only 67%.

This is not very good (the usual suspects — Fitbit, Apple, and Garmin — are better at around 80% accuracy, depending on the metric) but I also agree with this commenter’s view on HN: as long as you are using your fitness tracker mostly for motivation1 it does not really matter how accurate individual measurements are. It’s more important that overall trends can be checked. If, for example, the step count is systematically over- or underestimated you can still get much use out of tracking, because you can still check whether your new jogging schedule moves the needle as much as you think it does.2

It’s the measuring itself, and the difference you feel that is also backed up by data, that does most for my motivation.

  1. That’s my number one use case. 

  2. In terms of bias-variance trade-off, we may want to trade variance for bias in this use case.