Review: Jawbone UP

Three weeks ago, I bought a Jawbone UP. It’s a movement-tracking bracelet that can be plugged into your iPhone (or Android device) once or twice a day to transmit your activity to a mobile app. (You can also download your tracking data as a CSV spreadsheet by logging into the UP website and ‘crunch’ it in Excel if you like.) Web reviews of this device typically talk about how Jawbone originally released a highly flawed version that suffered from manufacturing issues and wasn’t properly waterproof which they eventually pulled from the market, embarking on a complete re-engineering exercise. The PR engine seems to have positioned this as a kind of laudable “they made things better for all of us” effort, and so every subsequent review of the device takes the re-launch as its starting point: is the re-designed UP band better than the original, flawed version?

Since I never used the original version, I’d like to look at it from a different perspective—on its own terms. Does it work? Does it do what it claims? Is it useful for me?

The bracelet comes in a variety of sizes. The retail boxes (in Canada, it’s only available at the Apple Store) come with a plastic sizing loop that can be used to determine the correct size for you. As in the picture above, a number of different colours are apparently available, but the Apple Store I visited only had the onyx (black) and light green versions available. UP is comfortable to slip on and wear, and the texture feels pleasant on your wrist. While the fact that the band’s ends aren’t joined makes it easy to put on and take off, it does tend to get caught when putting on long-sleeved clothing such as jackets or sweaters (the rubbery texture is just sticky enough against fabric).

At the core, UP has two different modes: activity tracking mode, and sleep tracking mode. UP can track step-based activity during the daytime and is able to measure the length and quality of sleep during the night. The sensor network inside the band needs to be switched from one mode into the other by means of a single button, which is at one end of the band (the covered audio plug used to charge and transmit data to the mobile phone is at the other). Before going to sleep, you press the button and UP goes into sleep tracking mode, briefly indicating its status by lighting up a built-in LED icon. After you wake up, switch it back into daytime tracking mode using the same button.

There are some auxiliary functions: UP has a built-in vibration alert that can be used as an alarm clock—UP will ‘gently’ shake you awake before your target wake-up time, while you’re sleeping lightly (it can track the difference between deep and light sleep). The vibration alert can also be used to remind you to get up and move during periods of inactivity during the day; you can set your preferred interval (i.e. “if I haven’t moved in 45 minutes since moving the last time, alert me”).

When you initially set up UP’s app, you enter a variety of personal parameters, such as your age, gender, weight, height etc. These are used to calculate—more or less accurately, of course—the distance you cover walking and the (estimated) calories you burn.

There’s also a stop watch function, which is a bit of a misnomer in my opinion: realistically, it is best used to calibrate UP for your specific body and stride length, not to time sports activities. In other words, if you’re about to embark on a (longer) walk, you can set an activity start marker (two presses of the button of different lengths—I found this difficult to get right, even after several attempts); turn it off after you’ve arrived at your destination. Using Google Maps, you can then trace the exact distance you walked and use this additional data to calibrate UP’s software to more correctly report the distance you cover each day by walking.

The app is used to configure your daily targets for walking and sleeping: I’ve set mine to a 10,000 step target, and 7 hours of sleep (I know myself well enough to know that the suggested 8-hour sleep target is just a pipe dream). You then plug UP into the audio port of your phone once or twice a day and it transfers the data logged in the meantime to the phone. The app then tells you how you did relative to your target. Some additional statistics are shown in the app itself—such as when you moved most during the day, or when you slept deeply versus lightly. The smallish charts cannot be zoomed and, as such, are probably not good enough for real data nerds. I would classify them as a kind of ‘ambient business intelligence’—more like an infographic, less like real data visualization.

Most other reviewers mention that their main point of criticism is that the band lacks Bluetooth data transfer capabilities—that you have to plug it into your phone. (And it is somewhat ironic that this is the latest offering from the company that made Bluetooth headsets de rigueur). However, I for one wouldn’t want to wear a Bluetooth device that closely to my body 24/7. I actually prefer this relatively radiation-free version, and I don’t have a problem plugging it in once or twice a day. (Incidentally, Jawbone provides a special USB charging cable for UP, and the internal battery seems to last somewhere between 8 and 10 days, more or less as advertised.)

For me, the main point of criticism is much more complicated: I don’t think UP is particularly accurate in terms of its movement tracking. The main issue seems to be that UP seems somewhat unable to distinguish between actual steps walked, and other movements. I have had several days where I’ve barely left the house but performed various activities that caused me to move my hands a lot, such as vacuuming, washing floors, etc. On those days, the band logged surprising numbers of ‘steps:’ on one particular day, I barely walked at all due to poor weather but cleaned the house instead, and had, apparently, completed 7,000 steps at the end of the day.

The main issue with this kind of ‘mis-tracking’ is, of course, that I may have moved my arm 7,000 times but not my body. This is a completely different kind of movement, and the underlying calculated metrics don’t apply anymore: distance covered and calories burned are now clearly incorrect.

Some reviewers pick up on this inaccuracy, but mostly in the context of distances reported. I think that the device is simply too sensitive and question its tracking accuracy overall. If it is able to misinterpret cleaning for walking, I wonder what it would capture if I sat in my car all day, moving my hands on the steering wheel. Or what would happen if I typed at my computer all day.

Unfortunately, this calls into question the overall accuracy of UP’s data tracking and reporting: if the metrics shown don’t correspond at all to my lived experience, I have to wonder what happened. This is particularly troublesome for the sleep tracking function: since I can’t know how deeply or lightly I slept, all I have to go on is UP’s data in the morning. I will concede that nights where UP shows that I didn’t wake up several times do feel like better sleep than nights where I did. Beyond that, I have no idea how to evaluate its sleep tracking.

Now for the important question: is UP useful to me, despite its flaws? I’ll guardedly say yes. It accurately reports ‘days with more activity’ versus ‘days with less.’ If we invented a fictional metric called step equivalent movements (SEMs) instead of using steps—and were willing to overlook the distance covered and calories burned that UP reports—then it would be a useful indicator of overall physical activity.

I can’t help but feel that UP needs a GPS in the device itself. If the data contained geographic location, it would be much easier to distinguish ‘walk’ from ‘yoga’ or ‘cleaning’ and report it accordingly. I’m not sure what the technical and/or size limitations are for including GPSes in a device this small, but it strikes me that this would need to be the next evolution of this kind of tracking technology.

In the meantime—and regardless of how well Jawbone re-engineered UP externally—it’s a flawed tracking device that does an okay job at providing me with visualizations of what I’d call ambient information: I very much need to apply my own intelligent filters to the data I’m shown. The movement tracked is not necessarily the movement performed. But knowing how much I slept last night is useful, as is knowing that I had more SEMs on some days than others.