Why we are asking you to use it

We could build the consumer app on assumptions. Every company does, and it is why so many health apps assume you own four devices and remember to charge all of them. Instead we are asking a handful of real people to use a real app on their real hardware, and there are five things we are paying attention to.

​

1. What people’s setups actually look like

The single biggest unknown. Which devices, in what combinations, connected which way. Whether people use the manufacturer’s app as well. What they do when two devices disagree. This directly shapes how the finished app introduces itself to a new user. If most people arrive with one watch and no idea what an API is, the onboarding we have sketched is wrong and we would rather find out now.

​

2. What the data is actually like

Wearable makers publish specifications. Specifications describe the best case. We need to know what genuinely arrives: how often, at what resolution, with what missing. Does your watch report respiratory rate every night or only sometimes? Does it give sleep stages or just “asleep”? Is skin temperature a number or a vague deviation? Some of this we can only learn from your device, which is why we ask you to run Discovery now and then.

​

3. How often wearables are really worn

Not “how often should they be worn”. How often they are. Rings come off for washing up. Watches die at 4pm. People stop wearing things on holiday. A model built on the assumption of unbroken nightly data falls over in the first week of real life, and we would rather it fell over here.

​

4. Whether our engines are right

Four things the Workbench computes need checking against real bodies: They are built on published science, but published science is built on population averages. Whether they behave sensibly on you, with your watch, on a week when you were ill and travelling, is a different question.

​

5. What people log by hand

The Workbench lets you record twelve things no wearable can see — weight, blood pressure, what you ate, how much you drank, whether you were ill. We want to know which of those people actually use. Our guess is that two or three get used constantly and the rest almost never. If that is right, the finished app should put those two or three within one tap and stop pretending the others matter equally.