Early mistakes

I quite like the observation made by Arthur Yusupov in his book "Build Up Your Chess". "Most mistakes occur in the early moves in variations" he says, pointing out that calculating long variations are useless if miss a strong reply from your opponent at move 1. To overcome this he suggests that you learn to calculate short variations accurately.
One way of doing this is to solve Checkmate in 2 problems. As much as the 'practical' chess player dislikes Mate in 2's for their contrived positions, by solving them you are forced to take into account every possible reply from your opponent.
Here is a Mate in 2 from Yusupov's book, which I set as a puzzle for a group of juniors. While it took me quite a while to find the key move (eventually realising that it defeated the most obvious Black defence), the group of juniors solved it quite quickly, and then complained that it was too easy!