Lowest average centipawn loss is good. 0 Is impossible, Chess champions often go below 10, in amateur play, acpl is often in the low 50s, chess engines such as Houdini, Komodo, Stockfish, AlphaZero and LeelaZero all have a low acpl (around 5). In long endgames(such as KNB vs K) is normal to have low acpl if in winning side. Acpl(Average Centipawns lost)is a stat generated using the computer powered analysis(on Lichess.org or Chess.com, both have these stats).
A low centipawn loss means you made fewer mistakes, which is obviously good. But on the other hand it's much easier for humans to make no mistakes in lifeless positions. So in seeking to optimize this value, that might thus suggest that you should play to kill every position as quickly as possible to keep the centipawn loss low. Of course that's quite silly.
Basically the positions where we make mistakes tend to also be the positions where our opponents can make mistakes. And by contrast the positions that we can play very optimally in also tend to be the positions that our opponents can play very optimally in.
So is it good or bad? If you look at the games of aggressive world champions you'll probably see a relatively high centipawn loss. If you look at the games of very defensive world champions you'll probably see a relatively low centipawn loss. In either case you'd probably see a lower centipawn loss than you would from a weaker player.
So I don't really think there's an answer. It's just a correlation where you're likely to do more harm than good trying to optimize for. It's like seeing that more rich people than poor drive Audis and so deciding to buy an Audi to increase your chances of becoming rich.