1. Look at all checks.
2. Undefended pieces.
3. Pins and forks.
4. Pieces (esp. the King) that do not have any “escape” squares.
5. Masked pieces (i.e. pieces on the same line)
6. Pieces that may be performing more than one defensive task
7. Finally, briefly look at bizarre and surprising moves, sacrifices, Pawn breaks, “obviously unplayable” moves.
Item 7 is what I used to find the winning idea in the following position after White’s 15.h4:
The game so far has revolved around Black’s attack on the White e-Pawn which he has adequately defended and his last move was played with the idea of forcing the N from g3 where it pressures the e-Pawn. A secondary objective was to commence his own attack against the Black K.
Let’s assume Black does nothing and makes a neutral move, say 15…Bd7. Then we have: 15...Bd7 16.h5 Ne7 17.h6 Nf5 18.Bf4 with equal chances for both sides.
Putting Item 7 into practice I looked at 15…Rxb2, but it doesn’t take long to realize that is a bad sacrifice. So, what about 15…Ba3? Then if 16.bxa3 Qe7 17.Qf3 Qxa3+ 18.Kd2 Rd8+ 19.Bd3 the attack is at a standstill and White is much better.
You’ll notice that what saves White is the intervention of his B to shield the K from check. This gave me an idea. As things stand now, if my Q were on e7 White couldn’t intervene with his B because the Q has it blocked; he would have to meet …Rd8+ with Qd3 and the Q would be lost.
This is where I got the idea for 15…Qe7. White should have played 16.Qf3 rendering the sacrifice on a3 no good. I had a suspicion he would never consider the sacrifice and would without hesitation play his intended 16.h5. That’s exactly what happened and the game continued from the diagram:
15…Qe7 16.h5? If 16.Qf3 Bd4 17.Bc4 is slightly better for Black. 16...Ba3! 17.bxa3 Other moves are no better 17...Qxa3+ 18.Kd2 Rd8+ 19.Ke3 Qxc3+ 20.Kf2 Rd2 and it was all over in a couple moves.