You're doing it almost right. Doing your own analysis first, and then running it through an engine. What you do need to change is all the labels: "mistake" "inaccuracy" "blunder" "best", etc.
The only thing you should be using an engine for is to check for blunders and missed tactics.
I have recently taken up the habit of analysing my games without the engine, I go through each move, writing down what I was thinking, what moves I was considering. After I have done this I check with the engine to see if my analysis was correct.
My issue is that sometimes what I thought was a mistake was the best move! And what I think is a great move is an inaccuracy or a mistake haha!
I then feel like I am not really learning or possibly even harming my game through this approach.
I will add a disclaimer that I cannot afford a coach. I am looking for any resources/books on analysis methods, help or any resources that might help me to learn more from my analysis.