My test of the Fitbit Air revealed the health tracker’s calorie counting errors – here’s why

Follow ZDNET: Add us as a favorite resource on Google.

Highlights taken by ZDNET

• I tested the Fitbit Air’s heart rate data.
• I compared it to the Polar H10 chest strap, a reliable heart rate monitor.
• The Fitbit Air is great for monitoring heart rate, but it’s not as accurate as a calorie counter.
  

  ---

How accurate is that health tracker on your wrist? The truth is, you don’t indeed know. The companies that create these devices equip them with sensors that detect every heart rate rise or dip, but their trackers are designed for casual, recreational use. Inaccurate things happen.

Still, many of the top health trackers out there are surprisingly accurate, and technological advances have brought each new generation a little closer to the gold standard for exercise and heart rate monitoring.

Also: Google’s Fitbit Air is a $99 screen wearable I can take seriously

I tested this on Fitbit Airone of the biggest releases of the year in the world of fitness. Using Polar H10 chest strap as a control, I recorded my gym workouts to see how the devices compare. While the Polar H10 chest strap is also a consumer product, it is generally considered the gold standard for heart rate monitoring for its technical accuracy. One study found that the chest strap showed “almost perfect agreement” with the reference Lead II ECG system, for example.

My exercise routine consisted of both strength training and treadmill sessions, my usual combination of weight training and cardio. Here is the breakdown.

On the treadmill

I started walking, checking heart rate measurements at that time on both Polar and the Google Health app. Switching between the two, the Fitbit Air was within a point or two of the Polar chest strap.

As I started running, the Polar quickly tracked this increase in heart rate as the Fitbit Air reading lagged behind. For example, when I started running, the Polar recorded a heart rate of 141. At the same moment, the Fitbit Air recorded a heart rate of 109. In the next minute, the Polar recorded 128 bpm, while the Fitbit recorded 112 bpm.

Also: I tracked 3,000 steps on my Apple Watch, Google Pixel, and Oura Ring – this one was very accurate.

After some calibration, the Fitbit Air caught up to my Polar chest strap’s heart rate reading. The devices showed the same heart rate data while maintaining a steady pace, but increasing or decreasing my pace caused some temporary discrepancies in the data.

This makes sense, since the Polar chest strap is closer to the heart than the Fitbit and uses electrodes to measure activity and heart rate, so those changes can be recorded immediately, while the Fitbit Air has to wait for that heart rate change to make its way from the heart to the wrist, creating a delay in reporting data.

Here are two heart rate graphs, showing the same metrics but slightly different timing, as mentioned above.

Heart rate metrics performed swimmingly with errors of less than 2.5%. The Fitbit Air underestimated the calories burned by about 12% during my treadmill session. These trackers calculate calories burned using a formula that includes heart rate, weight, age, gender, activity intensity, and more. It is not as simple as collecting heart rate data, so that errors can be compounded, as we will continue to see in my strength training data.

Strength training

To compare the accuracy of the Fitbit Air and the H10 chest strap, I used both devices to track my full-body workouts at the gym. These included lat pulldowns, dumbbell squats to shoulder presses, dumbbell lateral raises, stability ball exercises, and the hip abductor machine.

For both exercises, the smaller data points and heart rate are more accurate. Calories burned and heart rate greater than deviation from chest line, calories burned three times more accurate in my treadmill session.

Also: Whoop vs. Fitbit Air: I used both to track my health and fitness for a month – this one is better

Unlike the first test, the maximum heart rate decreased by 11% in this test. This may indicate that the Fitbit Air does not have short spikes in heart rate during short bursts of intensity, as my heart rate rises and falls between my workouts and breaks.

Because my strength training sessions involve a few short bursts of high intensity and then back down to base, the Fitbit Air is likely to miss the peaks of these heart rate spikes. A second experiment proves that the maximum heart rate is less reliable during periods of weight training than during periods of intensity, such as treadmill or cardio exercise.

An important point

I’ll start by disclaiming that two tests aren’t enough to draw firm conclusions about the overall accuracy of the Fitbit Air. This was a novice test at my local gym — not a lab test. However, this test helped me understand health trackers in a new way.

For example, health trackers worn on the wrist or finger will not capture rapid fluctuations in heart rate, due to their position on the body. They are far from the heart, so it takes a long time for blood to pump to these extremities and show the heart’s transient fluctuations.

This results in the device not always capturing subtle changes in heart rate or showing them in your workout statistics. Choose heart rate monitoring if that’s something you like during training.

Also: I’ve been wearing the Oura Ring 5 for a month, and two major improvements make it much better than the 4.

The Fitbit Air ended up recording more accurate heart rate data similar to chest strap graphs. Despite the 11% deviation of maximum heart rate strength training from the Polar chest strap, the device recorded heart rate data with low accuracy. I was very impressed with these datapoints and would recommend the Fitbit Air as a reliable health tracker for these reasons.

But with calorie tracking… that’s another story. It is already difficult to calculate calories burned, and each tracker will calculate it differently. I didn’t expect the Fitbit Air to drop 11% to 30% in its calories burned estimates. If you’re using a device for food or weight management, I’d recommend taking those data points as ballpark estimates rather than target points.