In a LiChess game (Vakeesh vs. becky82), during the game I thought I played well, and won as black without and real complications. Here's the game:
[fen ""] [Event "Rated Classical game"] [Date "2019.01.23"] [White "Vakeesh"] [Black "becky82"] [Result "0-1"] [UTCDate "2019.01.23"] 1. e4 c6 2. d4 d5 3. e5 Bf5 4. f4 e6 5. Be2 c5 6. c3 Nc6 7. Be3 Qb6 8. b3 cxd4 9. cxd4 Bb4+ 10. Nd2 Nxd4 11. Bxd4 Qxd4 12. Nf3 Bxd2+ 13. Nxd2 Ne7 14. Bb5+ Nc6 15. Rc1 O-O 16. Bxc6 bxc6 17. Rxc6 Qxf4 18. Qe2 Rfc8 19. g3 Qd4 20. Rxc8+ Rxc8 21. Kf1 Bd3 22. Qxd3 0-1
However, the post-game computer analysis gives me a "80 average centipawn loss", which implies that for my average move I blundered the equivalent of 4/5-ths of a pawn. (My opponent had "123 average centipawn loss", outblundering me.) I'm not sure what to think of this.
Question: What can I learn from my "80 average centipawn loss" game, which I thought I played well?
It seems I repeatedly played "a good move" rather than "the best move" (as there were often multiple winning moves).
Edit: I'm after humanly understandable "take home" lessons from the game, which I can apply to other games in general (when using an engine is considered cheating). I'm also after an evaluation of the applicability of the engine analysis (available from the linked game) to a human player trying to improve.
I'm not seeking "this move is better because the computer says so", I'm seeking how I can find the best moves as a human in a tournament game.