A session is good if it changes one future decision. That is a lower bar than most players think.
Players judge a session by the score. A good session is one with solved boards and high accuracy, a bad session is one with misses. This is almost backwards. Scores are the visible surface of the day; they say very little about whether anything was installed.
A better standard is a single question: what will I do differently next time? A session that leaves one concrete change is a good session, even if half the boards were failures. A session with perfect scores and no change is entertainment. There is nothing wrong with entertainment, but it should not be mistaken for training.
The change has to be concrete. "I should be more careful" is not a change. "I will name the ending before move three" is. A concrete change is checkable at the start of the next session, which is exactly what makes it usable.
This standard also fixes the session length problem. When the goal is one change, sessions end when the change appears rather than when the boards run out. Sometimes that is twenty minutes, sometimes five. The session is done when it has left something behind.
Keep the change on the note card and test it tomorrow. Click here for a fresh board when the next session starts, and the unlimited ladder for the extra volume a specific change usually needs to stick.
One change per session is also a realistic rate. Ten sessions with ten concrete changes is a different player; one session with ten changes is a confused one. The constraint is not modest. It is the mechanism that makes the change survive the next board. Write the change before closing the tab. A change recalled a day later has usually become a vague intention, and vague intentions do not survive the next hard board.