iPhone versus Fitbit HR - the myth of accuracy

[Paolo.c]

I have been concerned by the rather large differences noticed between my iPhone step-count and my Fitbit, where the Fitbit was seen to be undercounting. So, I worked this out today by going on a 3km brisk walk with tracking. Fitbit gave me 2,567 steps versus iPhone’s 3,469. That’s a whopping 26% difference!

Here is why

iPhone 5s (mine) currently has no way to add your stride length. My stride length for walking is 72cm and 90cm for running, as set on Fitbit. I’ve tried 10+ different apps with stride capability - all that seems to do is increase or decrease my total distance, with no visible effect on step count.

How Fitbit calculates a brisk walk

I walked slowly yesterday for 1km with my 4-year old son and was rewarded with 1600+ steps for my effort. So, by rights, today’s 3km should have been 4,800 steps! Correct…if I was walking at the same pace as yesterday, which I was not…

As mentioned above, I walked all 3km briskly. Doing a little maths, let us assume that my stride length was greater than walking pace but less than running pace. The difference between 72cm and 90cm is 18cm. Averaging it out gives me a 81cm brisk pace (somewhere between walking and running).

0.81 x 3000 = 2,430
Quite close to reality!
Percentage difference between 2,430 (my guess) and 2,567 (Fitbit’s guess):  5.34%

Why does this matter?

Simple - I am taking less steps. My stride is longer and therefore I’m covering more distance with less steps.

I have read post after post on here regarding the accuracy of Fitbit HR and I started to believe it until I worked it out. Fitbit is quite accurate. For those that believe the iPhone is king ‘because it uses GPS to calculate distance’ - this assumption is incorrect. Steps on an iPhone are calculated based on an accelerometer - same as my Fitbit, but minus my stride length. The GPS aspect only comes into play when you actually turn it on, perhaps by using a third-party app. But I’m guessing all iPhone users will sigh at this point, what with the very poor battery life in all models! 🙂

I’m a reformed couch potato! I feel better when I look at my iPhone step count and groan when looking at my Fitbit, but that just makes me work harder. After 3-weeks of Fitbit, I stopped getting hung up on the step count alone. What is important is the active time per day. Happy stepping! And I hope this helps to clear up some of the issues with accuracy.

[RobertoME]

Hello there @Paolo.c, I was reading your post and I must say is very impressive and accurate. Most of the mobile devices nowadays use GPS to track different type of activities, like Android or iOS. However the way the Fitbit works is very different as you explain here. Your tracker calculate distance by multiplying your walking steps and walking stride length. Similarly, your running steps are multiplied by your running stride length. Reason why not necessarily match a pre-measured distance, nor will it show exactly what a GPS device will show.

I will keep your post in mind for future reference and hope to see you around again.

[SkippySteve]

Paolo,

unfortunately your math is incorrect. 

If your stride length is 81cm

the correct equation would be 3000 divided by 0.81. Which would be 3703 steps. 

Cheers

[Iamlui541]

That’s incorrect. His stride is increasing, therefore, he covers more distance in less steps, so his steps must decrease not increase. .81 x 3000 is correct.

[Sensei-D]

You’re never going to get a device that’s completely accurate. My Fitbit added at least 30 steps while I was on a bumpy bus ride today.

[Iamlui541]

Only 30, you are lucky. My charge 2 adds between 500-1000 on the bus every morning.

[Sensei-D]

I have a versa, which I think has a more accurate tracker, although others might not agree.

[Mukluk4]

I don't understand quite what you are doing with your math...

If you are just trying to determine step count, why are you using distance/stride length?  Stride length can change with speed, variance in incline, or uneven ground.

The easiest way to see accuracy of step count is to either count in your head 100 steps and then see what your phone/fitbit has.  Or use a counter (such as a clicker) for a longer distance and then match to your phone/fitbit.

[Sensei-D]

It’s not quite that simple. In an ideal environmen at a consistent pace, it will always be accurate. However, if you’re moving slowly, where it doesn’t detect a jolt, it won’t count it and if you’re on a bumpy moving vehicle, it will think you’re walking and adds steps.

[MarcoGFitbit]

Hello everyone, I hope you're doing well. 

Thanks for bringing this to my attention. Please note that Fitbit devices calculate distance by multiplying your walking steps by your walking stride length and multiplying your running steps by your running stride length. We estimate your stride lengths using your height and gender. If you have a Fitbit Blaze, Fitbit Charge 2, Fitbit Ionic, or Fitbit Versa and track one or more runs using on-board or connected GPS, we evaluate that data to automatically update your running stride length. For best results, run at a comfortable pace for at least 10 minutes.

If you prefer to measure your walking and running stride lengths yourself and enter them manually, see How do I measure and adjust my stride length?

Note that when you're tracking an activity with GPS, Fitbit calculates your distance using GPS data rather than steps. If you begin moving before you get a GPS signal, the tracker will calculate distance using steps and stride length as described above until a GPS signal is found.

Fitbit devices have a finely tuned algorithm for step counting. The algorithm is designed to look for intensity and motion patterns that are most indicative of people walking and running.  When working at a desk, cooking, or doing other arm movements, a device on your wrist can pick up some extra steps if it thinks you're walking. Many of these situations—such as working or cooking—do include a few steps in-between stationary periods so the device tries to give you credit for those steps. For the vast majority of customers, the amount of stray steps accumulated is negligible when compared to the entire day. Your device has been tested against driving, buses, trains, and other public transportation and should not give you extra steps when you’re traveling under normal conditions. On bumpy roads or in a car with a stiff transmission you may get extra steps. Though a few extra steps won't affect your trends, it's possible to log a driving activity and remove the steps from your account totals. For more information see How do I edit or delete Fitbit data and activities?

I hope this resolves your inquiries, if there's anything else I can do for you, please feel free to reply. 

[CoriBukowski]

How does the above info apply to the iPhone app? Is it as accurate in the way it calculates movement and activity?

[MarcoGFitbit]

Hello @CoriBukowski, thanks for joining the Fitbit Community, it's great to have you on board. 

Thanks for participating in the Fitbit Community and for posting your question. Please note that your iPhone uses different sensors in it to calculate movement. These sensors provide information to the Apple Health and the Fitbit app. The main difference between both is that they use a different algorithm to calculate steps, distance and calories burned. Fitbit uses MobileTrack, which uses the sensors of your phone and the information might also differ from the one on your tracker depending on the type of movement made by your wrist and the placement of your phone. 

I hope this resolves your inquiry, feel free to reply in case you need anything else.  

[Rockgin]

I understand that the iPhone and Fitbit (I have a Charge2) use different sensors to measure activity. However, in my experience, I have found that the Fitbit over-counts steps/floors/distance. Perhaps I'm not fully understanding the settings correctly, but the Fitbit is adding significant "activity" when driving or taking the train - ie. non-walking/activity moments in time. It seems that the only way I can compensate for non-walking activity time is to manually input it into the log...which is a pain in the **ahem**.

Today, I drove and took the train and input 2 log entries into the Fitbit Activity dashboard which totaled about an hour's worth of non-walking activity. I had my Fitbit on and my iPhone in my pocket. This is the "count" for each as of about 1pm today:

iPhone: 

-- 5,81 steps

-- 2.1 miles walking/running distances

-- 15 floors

Fitbit:

-- 7,655 steps

-- 3.38 miles 

-- 23 floors

Thats a significant variance between the two. The iPhone's stats "feel" more accurate to me based on my gut feeling of what I've done today. It seems that Fitbit is counting any arm activity, driving/transportation activity as "steps" and mileage, and providing me with a higher estimate of my activity vs. what I'm actually doing. I also know that comparing data sets that have different sources is not a good comparison and that there will be variance, but in terms of accuracy, it's becoming very difficult for me to rely on the actual activity mileage that I'm accruing based on Fitbit data.

I'd love to know of a solution to the driving/train issue and any other comments relative to Fitbit accuracy. Thanks. 

[SunsetRunner]

I have the same issue with my iPhone vs Fitbit charge 3! My difference for today is about 5000 steps and 4km 

[Rockgin]

My solution: Pick the fitbit data or the iphone data and stick with it. There's no need to compare. Once you set your goals and reach your averages on one or the other, you can then work towards improvement or surpassing your averages. 

I now only pay attention to my Fitbit data, because I also like the sleep data. I know what my averages are after using it for over a month as well as my sleep averages and I now know what I need to do to maintain and surpass my daily, weekly, monthly averages.

[ChubbyDudley]

Rockgin. The fitbit logged 13X as many steps as iPhone. That's a HUGE discrepancy.  You said your activity was non-walking. Looks like the fitbit has a serious issue.

[Johnnirgee]

Today iphone app10,797 fitbit 14717 mowing grass walking something is full of...........