Another good example of where metric units don't work out so well is estimating distance by pacing. Most people don't have a yard or metre-long average stride. I can do metre-long strides, but I have to pace rather unnaturally and with very long steps. My average stride, and most people's actually, is closer to two feet. So is it easier to count paces and multiply by two, or count paces and multiply by 2 and divide by 3?
Which is the reason we had a pace-count course setup during land navigation exercices in ROTC. We'd follow a 100m course, count our paces, go back the other direction, count again, then average the two numbers. That was supposed to be able to allow us to convert our map distances into an easy to remember number when looking for waypoints. For the most part, it worked okay, since the pace course was a good average of the terrain. However, you'd still have to fudge a bit due to fatigue when your stride got shorter.
But in the end, it was still a LOT easier to use metric than imperial, simply because the course was probably 8-10 square miles. Imperial would have worked for road distances, but not when you're going through the brush trying to find specific points and cut-ins from the trails.