What accuracy % do you consider a “good” game?

a lot of end game moves that are bleedin obvious will boost up the accuracy score.  i agree with the 80% thing though, but there are a lot of extenuating circumstances.... at my level (high 1300s) i will simplify when  ahead due to trying to avoid blunders and while technically the accuracy might go down, the probability of a win goes up

I think at 1800-1900 a good accuracy, at least your goal is to get 95

Anything above 90%.

Below that, I feel disappointed in my play.

Though, mistakes also mean that there's something significant to learn from, so they're a blessing in disguise.

I think to feel good about the game report it depends not really on my accuracy but on my opponent's accuracy 

Ex: I don't feel accomplished for a win if my opponent blundered their queen on move 5 and got 2% while I got 99%

I feel more accomplished if me AND my opponent score a decent accuracy, say 80%+ and I only just outplay him

I am a 773 rated player right now and I just got a 88.9% accuratecy game and was wondering if that's good for my elo?

 
well here I was feeling great about my 83% accuracy game

85%+

I consider 80%+ as my good games. 

I'm 100 behind you in rating and I'm overjoyed if my accuracy is above 70!  

Just played a game with 89% accuracy [Faint!] and the analysis still made me out to be a chump. (The engine really didn't like one move in particular, but I still think it was a great move ... even AFTER the engine showed me what it thought was a better move.  "Oh well!")

I played a game with 99.9% accuracy.

Accuracy scores are strongly related to the length of the game. Short games tend to have higher scores if you win, and low scores if you lose, and also because theory boosts your initial accuracy. As games get longer and longer, you tend to collect many more goods instead of bests.

Also, a lot of times in completely winning games where no matter what you do you will win, you will find yourself losing accuracy. For instance, I was always told that if I was ahead, I should try to prevent counterplay first, whereas the computer will ding you because you didn't go for the fastest mate. Also, if you're ahead a couple of pieces and just decide to trade down, the computer will often ding you for inaccuracies. You're doing what your coach always told you to do, winning is winning, but the computer doesn't like it. Poo on you, Mr. Stockfish.

Chess.com's accuracy stat isn't really a useful metric.

For me, it can't really be considered a good game unless the opponent put up strong resistance. A walk-over is not a "good" game.

For a game that is at least 25 moves long where most of the moves were not forced, I would have to say that 90%+ accuracy (expert strength or better) is generally an indicator of a good game.  

My overall accuracy is 58.1 but that includes games from when I was starting at 500, so. . .

My accuracy in the last 30 days or so is around 72%.  I would say anything above 80 on a game that goes deep is pretty good for me!  

Sometimes I get games as high as 90s - but typically it would be short games where an opponent blunders a piece and I'm able to pounce and put them in a bad spot.

But like - I've had excellent games where I won decisively when accuracy is in the 60s - so unless you're against a computer I don't think accuracy is the end all and be all. . .

As NikkiLikeChikki pointed out, the moves that a computer considers accurate are not necessarily the moves that a human player would regard as accurate.

In particular, once you've gained a winning advantage, the most important priority (for human players) is to prevent your opponent's potential counter-play. A computer would not consider that to be accurate, since the computer is programmed to favor the mathematically quickest win.

The first thing to discuss is what exactly is meant by "good". It is a pretty broad term after all. Calling a game "good" may equate to calling it "instructive", "precisely played", "exciting", "innovative", etc; or any combinations of these. Not everything that can be "good" about the game is included in accuracy.

accuracy isn't accurate