We also really stand behind that 20% number. We brought 4 professional tailors in and 35 people, and we were 20% more accurate than the professional tailors. If anyone wants, I can go into more detail on the calculations of us vs. tailors.
We had each tailor measure each person across 17 different measurements. Across those 4 tailors (on each person and each measurement), you could see a standard deviation. So then we also looked at the standard deviation of the average of those 4 tailors vs. the MTailor software's system. Our standard deviation was 20% lower.
Happy to go into even more detail if people are interested!
You're definitely right about precision vs. accuracy. The phrase never polled as well. Maybe if we'd run our surveys in the statistics department it would have been different. :-)
You can't wear a normal, loose t-shirt or shirt; that unfortunately destroys our accuracy.
Yes, the Kinect would probably be more accurate (it was one of our original ideas), but it wouldn't have made a very good business. No one would have been able to easily use it in his home (not many people own a Kinect, 3rd party apps for xbox aren't big and when I have seen a Kinect, it doesn't capture your whole body).
We found that our accuracy is more than sufficient for shirts (we've shipped a bunch already and gotten overwhelmingly positive feedback). Accuracy is just 1 aspect of a great looking shirt though; style is also really important, and we've spent a long time refining that as well.
https://www.flickr.com/photos/kylemcdonald/9150428195/in/poo...