I was doing an analysis in AI sensei (AI Sensei | News) and really like the categorization of the moves (e.g. blunder, mistake, etc.) this is something that chess.com does as well and it helps to quickly understand where you went wrong. Generally seeing the change in score (e.g -0.3, +1, etc) is fine but not as clear as showing “mistake” or “blunder”.
If this is not something that’s available right now (perhaps with AI review?), I think it should be.
If Im not mistaken, AI Sensei also just looks at the point loss and bases the categories on them. Like for example everything in between a -2 to -4.9 point loss would be an inaccurracy, -5 to -9.9 a mistake etc. - depending on your set rank.
So, you can make your own margins and have the same thing, minus the cool looking visuals - admitted 
And besides that, sometimes a -5 point move may be the best move to play because it simplifies a won position significantly and increases the winning chances. On the other hand, some -1 point moves can be pretty stupid and dont get pointed out(even get rated as good moves).
I think every -x point move may need its own interpretation if its a mistake, blunder, inaccuracy or whatever.
I think those are some good and valid points. There could be improvements there overtime but perhaps an initial feature that could be a toggle that allows users to turn on move categorization. A tooltip could be provided with the caveat that it may over generalize moves. Even with Chess.com it’s not perfect and it categorizes moves in a strange way but that’s okay.
If there’s not a need for this then no need to implement it. But if there are others that are interested, then this seems feasible to implement.
I agree that interpretation of moves is more nuanced than just this AI evaluation, but it’s also worth something to have a suggestion from it. I found this feature on chess.com very useful so maybe others do too.
Considering how difficult it is to categorize a move like you hope to get, I would rather keep the raw data and mix my own opinion with it to give myself a category for each move.
I don’t use c.c, but I play on lichess occasionally, and I think the big valueadd of identifying “blunders” there is the review with ai option where it shows you positions where you make a blunder, and asks you to find a better move. If you try something it wasn’t thinking of (I assume “had less than some threshold of playouts” is a rough equivalent), it analyzes your try, and if it turns out to be not that much worse than what it wanted, it marks it correct
What OGS already does with the red dots is similar in concept