Learn more from your data

“I have lots of data from my Oura ring or Apple Watch, but I’m not learning anything that matters.”

This is one of the most common barriers people run into in their personal science projects. This barrier is not specific to Oura or the Apple Watch, but we hear the complaint most about these two wearables because they are so popular and so genuinely good at what they do. That is, they deliver mostly reliable data on heart rate, movement, and sleep. There are a lot of details about how to make this data more trustworthy and compensate for some of its known weaknesses. But the question addresses something more important than that: What are they even supposed to be doing for me?

The reason many people get less than they hoped for is not the quality of the sensors or algorithms. It’s that these devices arrive with the meaning already packaged in. The app hands you scores, trends, visualizations, and even coaching advice. All of it is built around topics the designers anticipated: readiness, recovery, sleep quality. If your question closely matches what was imagined when the products shipped, you’re in luck. You know what you are getting, and you can see its value. But if your question is even slightly different, or if you don’t know your question yet, all the data, no matter its quality, is going to seem irrelevant.

Here is our best tip for getting more out of your data:

Use the foreground/background pattern.

Your foreground observations will focus on the phenomenon you care about most. This is the core of your question. If you are tracking sleep, there is probably a reason you want to know about your sleep. Is it because you are tired during the day and want to have more energy? Is it because you are training for a marathon and are pacing runs and recovery? Is it because your partner says you snore? There are infinitely many questions related to sleep, and you will get more out of your data if you can think a bit about what you’d really like to know the most. This is not a permanent decision, obviously. You can do a series of projects that last anywhere from a day to a decade. But at the start, it’s really worth thinking about what you would be very glad to know in the future that you don’t know now. Maybe pick something you can find out about quickly so you are rewarded for your effort without delay. Once you have your question, then you can think about what you’d like to observe that is directly relevant. Degree of tiredness at 10 a.m. Snoring episodes, recorded on your audio-powered snore tracking app. Satisfaction level of your daily run. These are just a few examples. You can easily think of your own. The point is that once you have a refined question and a foreground observation, you are more likely to learn something that matters.

Now that you have your foreground observation, your data is going to provide extremely useful background. Background observations serve to explain variation in your foreground observations. So if you are especially tired at 10 a.m. one week, you can look at your sleep data to see what’s changed. The relation between foreground and background observations is especially interesting when you’re surprised; for instance, if you are unusually tired but you show consistent sleep. Why? The answer might turn out to be important! In looking at your foreground observations and seeing if they are connected to the most obvious background signals, you’re challenging yourself to think more deeply about what’s really happening in your life. That matters.

Finding a foreground observation is the single most important thing you can do to get more from your data.