Here is an example
https://www.chessgames.com/perl/chessgame?gid=1072945
White won, hence black must have made a mistake.
Tracking backwards from the final position reveals the mistake: 16...Be6? was the losing move. Black had to overprotect his vital central pawn e5 with 16...f6 keeping equality.
Can someone please look at my last Blitz game and explain to me how I was told that I had lost the game when my clock said I had 13.6 seconds left?
Without the game link i won't go through the big list, but sounds you maybe lagged or disconnected
https://support.chess.com/article/213-how-do-i-fix-my-disconnect-lag-issues
Here is one of my games. Not sure if it is what you were looking for but it's worth a shot
What you can do is this.
Remember after the game what you were thinking, some other candidate moves in certain positions. Make a note of a variation you were thinking about and write it down (AKA type in the moves).
Write down some of your thoughts (something like: I wanted to push the b pawn here, but that would weaken c4 so I decided to play this instead, and stuff that you think is relevant.
So you should annotate plus type in variations as well.
Now, if you don't use the engine, you will have a harder time checking out if your thoughts are correct. It is not impossible, but I would turn the engine on at some point to do many different things. But if you don't want to, then just try to check your observations just by looking at the position for some time. If you are uninterested in engine checking, then don't read further.
How can you use the engine?
I would check your own observations with the engine for sure, and some shifts in the evaluation, to see if you can figure out the reason behind those shifts. If you can't, just move right along, don't write down every engine line, only what you can explain. Sometimes my candidate moves were better than what I played in the game because of a specific reason.
Perhaps in the game you didn't have a plan. Try to make some stupid waiting moves for the opponent and try to figure out what the engine is saying. For instance I played one game a few days ago, and I was thinking about the f pawn push to open up the position, but there was other plan as well, where I could first push the a pawn to stop the opponent's play on the queenside and only then push the f pawn. Sadly I went with the third option. 
Oh, and spend a lot of time in the endgame.
The more time you use, the better in general, but if the game isn't that complicated, don't waste too much time.
I have a blog on this, but it is beginner oriented, and annotations are a bit long just to help them understand some positions, so it will not be that useful to you, 1 700+ rated player.
In general, my analysis is in Serbian, so I don't have an example for you.
I hope that those general ideas will be of some use to you.
I think it helps to use the computer but only after looking at the moves that lead up to the position slipping away from you for example. But I usually just like to look at the eval bar itself and see where for example you lost the advantage and see if the move you made was something that for whatever reason you intuitively think is normal at first glance then use the engine to show you the line of why it was a mistake/blunder whatever that way its more into your memory. Similar to how you would study checkmate patterns.
One last thing I'd add is one thing I've started to do lately that I've found helpful is whenever I'm in a position where I feel like I literally cannot find a move I'll go back and analyze that move and see what moves engine suggested at that time.
My first analyzed game.
With thanks to GM Pogonina (WGM, if it matters..), I've read both articles:
https://www.chess.com/article/view/how-to-analyze-chess-games
https://www.chess.com/article/view/game-analysis
With apology to @anaxagoras , who said not to publish won games. I won at the end, the middlegame was, as usual - terrible - it's not like I crushed him or anything.
I don't know what this analysis has given me so much, but it must have given Something.
Conclusion: Don't play h3 against just about anything, and be more aware of the 'sac for 2 pawns' by opponent - it may come quicker than you imagine. pay more attention to opponent's bishop moves?
Basically, I have violated a K.I.D rule - you should only attack on the flank when the middle is closed, but, since I don't know the 'why' behind this rule, and since I don't know to do anything else against it [but mostly, since I forgot the rule], I just attacked.
Homework: Learn what to play against a benonian-King's Indian Defense.
---
[Okay, so it's the conclusion after the analysis that helps you, not the analysis itself]
We all say that in order to improve, you need to analyze your games.
How?
Please, show one example. High theoretical explanations don't help.
I mean an example that teaches how to actually do it - not some magical annotation.
Preferably without a computer. - computers are terrible, and chess is 1500 years old.