I did a comparison of driving duration using Graphhopper and Google Maps Directions API. I compared about 4000 driving durations in Germany and found out that Graphhopper underestimates driving duration compared to Google ~20%.
Have you guys done any performance comparison on that side and checked how exact the calculated driving durations of Graphhopper are in reality?
Underestimated driving times is a known issue within cities and we are working on it.
Please see this new issue: Improved Feature: Better ETA values