Improvement is cumulative only if your work is stored. If every analysis session starts from zero, you are paying the same tuition repeatedly.
Keep every serious game
Played games should be stored automatically. Uploaded games should be imported once and tagged. The goal is a searchable record of the positions you actually play, not a museum of famous games.
Annotate the turning points
You do not need to annotate every move. Mark the first turning point, the critical calculation and the moment you ran out of plan. A short note is more useful than a long engine dump.
Separate the three collections
- My games: reviewed, annotated, with generated puzzles.
- Opening files: the lines and structures you actually play.
- Study positions: endgames, tactics and plans you want to revisit.
ChessUrge's game library covers the first collection. PGN export lets you move games into a desktop database if you prefer. The important thing is that the files are yours.
Review on a schedule
A library is only useful if you revisit it. Once a week, open one old game and one generated puzzle. Once a month, review your opening files. The schedule matters more than the size of the collection.
Do not hoard
A thousand unread games are not a study library. Fifty reviewed games with notes are. Keep the games that taught you something and let the rest be archive material.
See the whole game, not just the blunder.
Choose a review depth, see the maximum cost, and get move-by-move accuracy with blunder, mistake and inaccuracy markers.