Workout app and MapMyRun Distances Are Inconsistent

What's problematic with this is apps like RunKeeper give you an accurate GPS distance from the same iPhone that the Workout app uses to calculate distance.

Workout app is not using (or so how I interpret Apples language) GPS to calculate data as I mentioned. Its just using it as a data point to refine your stride length which in turn helps better approximate your distance. So they are not using the same data. As someone pointed out, perhaps on outdoor cycling, it is using GPS since the accelerometer would not be very useful.

I will say that Runkeeper should not be viewed as an official source of GPS data and thus being accurate. I have done 5 tests using Runkeeper, Runmeter, and AW compared against the actual mileage of my tracks using the Footpath app. Whats nice about Footpath compared to Google Maps is that you can draw your paths without having to be straight lines. So if you know that you crossed on a diagonal, you can map it a bit more accurately. I am in NYC and so what problem with GPS is that its harder to get a consistent signal because it bounced off building and or tree cover. I could show you the maps but basically you can observe these problems (which are also related to the GPS capabilities of the iPhone) in how the maps for both RK and RM show me crossing the street or doubling back when in fact i ran in a straight line up one of the avenues. Or on a park run, you can see the GPS jump to the wrong trail for a small portion. In total, these small inconsistencies can add up to a material difference; however other times, the impact is not as great and I will get readings within 0.02 for all three when compared against Footpath.

From my experience so far, I have amassed 9 runs and 15 walks with the watch all using the workout app and the phone. My last 5 activities have been very very close to the Footpath distance whereas the first 5, on avg were off by 0.4. These are all across very similar routes, with 2 outliers when i traveled. I am now about to test how these compare when no phone is being used, more for intellectual curiosity as in reality, i almost always carry my phone with me.

Okay but when you're running on very well-defined running trails with no deviation from the path, and RunKeeper agrees with the trail distance but the Workout app is off by 10% or more, I'd trust RunKeeper over the Workout app. Similarly, when I run in an officially-measured race and RunKeeper agrees with the distance & pace, but Workout doesn't, I'd say there's an issue with the Workout app.

I get what you're saying about Workout using GPS as an additional data point to calculate distance (combined with accelerometer data) rather than as the absolute truth, but if that causes it to underestimate distances so much, then I'd consider that a major defect in the algorithm.

I use Strava and the watch for cycling, the distances are reported the same.

Your issue has to be as a result of something else.

It has already been theorized in this thread and elsewhere that this mainly affects running. My walking distances have been spot on with the Workout app.

I didnt say trust AW over runkeeper, only that there are many variables that you should look at to ensure your perception of accuracy is correct.

And just my two cents but the AW should be thought of like a fitbit or any other device that is measuring distances based on steps (and ones doing so via your wrist). They are never going to be as accurate as GPS-based devices (ie a garmin watch or iPhone app using GPS). Until Apple either puts a GPS in the watch or decides that it will use actual GPS data from phone to measure distances when selected, you should not expect similar accuracy (although, i am not 100% on what is actually going on since the language is a bit ambiguous -- see below). Bottom line is that if you need distance accuracy down to the hundredth or even 10th of a mile, shouldnt be using the AW over something like runkeeper if both are readily available.

Calibrating your Apple Watch for improved Workout and Activity accuracy - Apple Support

Based on other threads i have seen, i think choosing Outdoor Cycle in fact only use the GPS to report which is why it is accurate for you. For runs and walks, their support thread seems to indicate they do not report GPS but only use it to make the accelerometer smarter. I guess an easy test (although have no idea what they have calibrated for outdoor cycle) would be to run but choosing Outdoor Cycle to see if the distance measurements are the same.

Firstly, says who? Apple Watch measures distances down to the hundredth of a mile/kilometer and lets you set goals down to the hundredth of a mile/kilometer, yet you're not supposed to expect hundredth-of-a-mile accuracy?

Secondly, people have reported discrepancies that exceed a quarter mile.

Lastly, calibrating the sensors uses the GPS to allow the Watch to better estimate your distance traveled when not carrying your phone. This is done by taking your phone with you on outdoor activities. The issue being discussed here is with the distance being wrong while carrying the phone. Obviously if that distance measured is wrong while calibrating (which is what an outdoor activity while carrying your phone is), then calibration will end up being incorrect as well.

My walks are accurate too, I don't run so can't test but if running and walking were treated differently (other than for calorie burn) it would be a little bizarre. It's quite clear from Apple's literature that GPS isn't limited to cycling.

The Apple Watch is obviously using GPS data from the phone when in outdoor cycling mode (and it's pretty accurate for me), so I think it's reasonable to expect similar accuracy for an outdoor run with the phone. The fault isn't our expectations, it's in the implementation of the outdoor running mode (with paired phone). Given how they're marketing the device ("a smarter way to look at fitness," "an advanced sports watch), it's definitely a huge flaw when the watch is showing your pace as 2 minutes slower per mile than it actually is, and assigning calories too low because of it.

This really shouldn't be that hard to fix, so I look forward to updated software.

I certainly have no idea how it all works but if they werent attempting to treat it differently then why have a separate activity for it? A plausible answer could be some analytics applied to stride length in terms of expected strides based on height/gender and error correcting based on the individual. For example, fitbit lets you set these manually -- walking stride length and running stride length.

I havent seen any literature related to cycling but for walking running, they are very specific about using the GPS to calibrate the accelerometer vs using the GPS to measure your distance. For cycling, not sure you can do anything but use the GPS.

Why have a separate activity. That's obvious, you burn more calories running the same distance as walking it.

I don't see why your suggestion is plausible, it uses GPS outdoors, not your strides (unless you don't have your phone), but then that's not the situation being discussed here.

Calibration is for measuring your stride to use indoors, not outdoors unless you have no phone, Calibration has no relevance in this discussion.

You can believe what you want, but they dont market it as a GPS watch and thus there is nothing to fix unless you submit feedback (which i have done). And they are very specific (or in interpret it as such) about whats it doing when you walk / run and that is to use GPS to calibrate the accelerometer. And thus you are getting steps based distance and pace, not GPS-based. Submit feedback on it and in the interim choose Outdoor Cycle when you run. I assume it will just report the GPS data which is what you want. But I assume there is no pace associated with cycling so that will be the compromise.

I agree that the different calculations for stride length, etc. for walking vs. running make complete sense -- when you don't have the phone with you. Why would you have the watch throw out perfectly good GPS data in favor of less reliable stride data based on generalized calculations? I suppose you could argue that you don't know if the GPS data is necessarily "perfectly good," but it seems to be much better than the stride data for many of us.

I've only had the watch for 2 weeks, so I'll be interested to see if the run accuracy improves over time. Maybe if the GPS consistently tells a different story than the stride data, the stride data will correct itself over time. Just hypothesizing/hoping.