Isn't 120 rides too few to come to a conclusion? Specially with different routes, different conditions?

It would have been better to see if they all told to take the same route and then compare the predicted time as well as actual time. It's possible that it started raining in one of the case or there was an accident, which added a lot of time to the predicted time.

FWIW, I use Google Maps and Waze alternately. I always arrive a minute or two earlier than the GMaps predictions. It's easy to note this from the summary page after you complete the ride. Waze, sadly, doesn't provide any summary. No idea about Apple Maps as I use Android.

Disc: Googler.