Do nothing without examining any checks you can give; and if you have active knights, do nothing without inspecting their prospects for forks. In this case either inquiry leads you to Nxe6+—a fork of Black’s king, queen, and rook. The only trouble is that the forking square is protected by Black’s bishop on c8. The bishop could be taken out first by your d6 knight, but then Black plays QxN and nothing has been accomplished; now his queen guards e6. So move to the next question: does the c8 bishop protect something else you can take? Its only other protectorate is the pawn on b7, but don’t be shy about considering a capture of it. The piece White has for the purpose is his other knight, and he sees that Nd6xb7 is another fork, this time of Black’s queen and rook. So this is what he plays, and it wins the exchange (plus a pawn) after Black ushers his queen to safety on e7. Black can’t play BxN because he needs his bishop to keep guarding e6 against White's worse forking threat.
The Black bishop was overworked. It protected two different forking squares. In retrospect you therefore can see that the presence of Black’s pawn on b7 was unimportant; what made b7 significant was that a fork was available from there, and this would have been no less true—though it might have been a little easier to overlook—if b7 had been empty. The point to take away is that a piece can be overworked just as easily by having responsibility for too many forking squares—empty or occupied—as it can be when it guards too many mating squares or too many pieces or any combination of these things. So while your tactical investigations can begin with a look at any pieces you can capture, they also need to include high awareness of any forks or mating possibilities and how the enemy fends them off.