Shouldn’t Fitness Trackers Be Accurate?

314xSphvUfLI’ve written about fitness trackers here several times. I currently have a Fitbit Charge, I’ve reviewed the Fitbit Flex, Fitbit One and Jawbone UP24, and I tried the Nike+ FuelBand.

One problem with these devices – aside from comfort, which is not always a given – is their accuracy. I’m surprised that most websites that review these devices don’t even discuss accuracy, when it’s not that hard to compare different fitness trackers, and determine how accurate they are. It’s almost as though reviewers expect them to be inaccurate, and don’t think that accuracy is important. (One notable exception is this Mashable review.)

I’ve been using the Fitbit Charge for a few weeks, and I’m surprised at how inaccurate it is. Fitbit, on their website, tout the accuracy of the device, saying:

“Charge has been tested extensively against our clip-based devices like the Fitbit One and Fitbit Zip. That said, because Charge is specifically designed for your wrist, if you move your body a lot and not your arms (or vice versa), you may get a slight difference in activity than you would see on your clip-based trackers. Additionally, since you’re more likely to wear Charge 24/7, you may count a few more steps.

“This is no different than any wrist-based tracker on the market. For most people, there may be no difference at all between clip and wrist based trackers or it may be within a few percentage points difference. That said, if you have a lifestyle where you move your hands a lot such as playing the drums every day, you may see a few extra steps on your Charge because we do want to give you credit for this activity.”

Yet it’s horrible. Compared to the Fitbit One, the Charge records 30-50% more steps, and, therefore, longer distance and more calories burned (and its “active minutes” is pretty wonky too). It even records some steps while I sleep. Granted, the Fitbit Flex records lots of steps when you drive, so I guess this is an improvement. And the Jawbone UP24 doesn’t record steps when you walk on a treadmill, so all the Fitbit devices win on that front.

Here’s another example of the Fitbit Charge getting it wrong. I went out to mail some letters today; the mailbox is 400m away. First, when putting on and tying my shoes, the Charge recorded 20 steps. Then, I set the Fitbit app to record my walk. When you do this, it uses the GPS in the iPhone (or other phone) to record a map of your walk, providing you with distance, time and pace per mile or kilometer. The app recorded this correctly, and said I had walked 794 steps; the Charge recorded 943 steps.

Most people don’t compare the accuracy of different devices, so they never know how inaccurate their devices may be. Precise accuracy isn’t essential; as I said in my review of three of these devices:

“Fitness trackers are motivators. While, on the surface, they claim to record data about your activity, the real reason people buy them is to motivate themselves to be more active. None of them are perfectly accurate, and they all have drawbacks.”

Thinking about this made me wonder how companies think they can sell these devices with such glaring inaccuracy. I understand the technical hurdles in counting steps using a wrist-worn device; but what Fitbit says, above, is that it’s not only possible, but accurate. They say that the charge “may count a few more steps,” but they’re disingenuous if they really think that it’s only a few steps.

Fitbit says:

“Fitbit is dedicated to developing the most accurate activity trackers on the market. Our team performed multiple internal studies to rigorously test the accuracy of our trackers. Through our testing, we have confirmed that our trackers are some of the most accurate wireless tracking devices.”

And, on another page:

“We’ve tuned the accuracy of the Fitbit tracker step counting functionality over hundreds of tests with multiple body types. All Fitbit trackers should be 95-97% accurate for step counting when worn as recommended.”

So the problem is one of a marketing message that is simply not true; where I grew up, we called that “lying.” 95-97% accurate would be great, if it were true. Assuming that the Fitbit One is the most accurate of the devices I’ve tested, because it’s worn on your belt, like a standard pedometer, none of the other devices are anywhere near that accuracy. The Nike+ FuelBand cops out, most likely because the company realized that accuracy is simply not possible with a wrist-worn device; they came up with “Nike+ Fuel,” a points-based system that has no relationship to steps. In a way, they’re the most honest; they don’t pretend to be accurate. However, Nike still claims that their software can count calories using an algorithm based on the energy you expend when you move. That’s pretty vague, and deflects the issue.

The fitness tracker sector is expanding rapidly, and the one thing companies should do is ensure that their devices are accurate. I think these devices do have a future, but only if we can take them seriously. For now, we simply cannot.