For me, accuracy is much better under the foot than on my laces

Question

Just thought I'd pass on my experience...
I did about 30 miles of running with the sensor tucked under the laces of my Asics (held in custom-made duct tape pouch -- who needs neoprene!) but was pretty disappointed at the wide range of inaccurate paces and distances it reported. It was especially off when I did some trail running on soft surfaces.
Then I got brave and took an X-acto knife to my Asics insole, and cut out a little cubby for the sensor, just under my arch, in the same location the Nike diagram shows it. And after another 15 miles or so, the results are FAR more accurate -- agreeing with mapmyrun.com down to the 0.1 mile on a 5 mile run -- certainly as accurately as I click on their map.
My guess, based on the Apple statement that the sensor is a piezo accelerometer that works by time the foot is on the ground, is that the accuracy is improved because (a) the sensor is horizontal under the foot, compared to ~30 degrees when on top of the foot, so when mounted on top, the vertical acceleration is confounded by some component of horizontal acceleration, and (b) there is some additional flexing of the foot that would be seen as a "muddying" of the acceleration signal when on top, whereas under the foot the 0-acceleration "on the ground" time is probably more clearly delineated, and thus easier to signal detect. (This would seem to fit the observation that the pace computation is way off when on soft dirt & sand compared to pavement, too). Just my engineering guesses, of course. But the improvement in accuracy so far seems well worth cutting a hole into my insole.

Answer 1

Thanks for the info. I had actually read about your modification when you first posted it. Great information. I am hesitant to cut into my two pairs of Asics GT-2110 as these have become my favorite shoe and probably the only shoe I will ever buy in the future. So I plan to use your velco under the laces modification on those.
I do have a pair of Saucony Omni 5's that I do not like as much. So I decided to try the cutting method on those. It actually worked very well. The sensor was very accurate and consistent. I could not feel it at all in my arch. Of course, the real test will be repeated runs to see if my foot develops any soreness.
I have also used a shoe pouch. I do not recommend this method. The accuracy greatly decreases and worse, it is much less consistent. I usually run around a 9 minute pace and using the shoe pouch method, the pace is always flucuating from like 7:00 to 13:00 even though most of the time it does settle around 9:00. I assume this is because the sensor is moving around too much.
I may try the marware product when it comes out. It looks like there will not be much movement with that one. And since I rotate shoes, it would be slightly more convenient than the velco method. But it would have to be accurate and consistent to make it worthwhile.
All in all I love this product. I refuse to buy Nike shoes. So I am glad I have found a few proven methods to make this thing work.