The short answer: stop trying to record the session and start recording four things: who was in it, what was promised, what changed in the world, and what the party still does not know.
Everything else is either in your head already or was never going to matter. A campaign does not collapse because you forgot a description; it collapses because a thread got dropped and nobody noticed for eleven weeks.
The break tends to arrive somewhere around session twelve to twenty, and it is not really about volume. It is about distance. Early on, everything that has ever happened took place within the last month, so it is all still loaded. By session fifteen there are events a year old that the players remember vividly and you do not, because they only had to remember their own character's version and you had to remember everyone's.
The four things worth keeping
1. Who was in it
A list of every named person and place that appeared, per session. Not descriptions, just names against session numbers.
This is the highest-value note in the whole campaign and the one people skip, because it looks trivial while you are writing it. It is what lets you answer "have they met this person before?" instantly, and that single question is behind most of the moments where a world stops feeling solid.
2. What was promised
Every commitment, in either direction. The party swore to return with the body. An NPC said the gate would be open. You said, out loud, that the caravan would be there in three days.
Players remember these with total clarity, much better than they remember plot, because a promise is a thing they did. A dropped promise is the fastest way to teach a table that the world does not really react to them.
3. What changed in the world
Not what happened. What is now different. The bridge is down. The duke knows their names. The shrine is empty and someone will notice.
This is the difference between a campaign and a series of adventures. It is also small: most sessions change two or three facts about the world, and writing those two or three lines is a minute of work that pays out for a year.
4. What the party still does not know
The one every note-taking system misses. You are holding a set of secrets, and the thing you actually lose track of is which ones have landed: what has been revealed, what was hinted at and missed, what they have quietly worked out and you have not noticed.
Keep a running list of open secrets with a note of whether it has been shown, hinted, or is still entirely yours. Reading that list before a session does more for continuity than re-reading a recap.
What to stop writing down
- Blow-by-blow accounts of combat. Nobody has ever re-read one. Who won and what it cost is enough.
- Dialogue, verbatim. Unless it was a promise or a reveal, the exact words do not matter.
- Descriptions of places the party passed through. If they come back you can describe it again, and if it mattered you will remember.
- Your own plans, in the session log. Plans belong with prep. Mixing what you intended into the record of what happened is how you end up remembering the version you wrote instead of the one you played.
The organising problem nobody warns you about
Most DMs who keep notes keep them chronologically, a document per session, in order. That matches how you write them and not at all how you read them.
Nobody ever thinks "what happened in session nine?" They think "what do we know about the Ashen Covenant?", and answering that from a chronological log means opening fifteen documents and reading all of them.
So notes need to be organised by thing, not by date: a page per person, place and faction, each accumulating what happened to it over time. Session logs are the raw material; the per-entity pages are what you actually use. If you are keeping only one, keep the second.
Whatever you use, it has to survive being boring
Every system works in month one. The question is whether you will still be doing it in month nine, at eleven at night, after a four-hour session, when you are tired and the group has just left.
That is the real constraint, and it is why elaborate wikis usually fail while a scruffy text file often does not. Pick the thing you will still do when you cannot be bothered.
The recap is the load-bearing habit
If you only build one habit, make it a short recap written the same night: the four things above, ten minutes, while it is still in your head. Not the next morning: the drop-off between "that night" and "the next day" is enormous, and by the following week you will remember the fight and not who was in it.
It also solves the other end. A recap you can read at the start of prep turns "what were we doing?" from twenty minutes of archaeology into two minutes of reading, which is where most of the thirty-minute prep method comes from.
Or don't write it at all
The honest problem with everything above is that it is work, at the exact moment of the week when you have the least appetite for any. Most DMs know what they should be recording. Very few are doing it at midnight on a Tuesday.
Which is why the other approach is to capture the session itself and let the notes come out of it. The names, the promises, the changes, all extracted from what was actually said rather than from what you can still recall afterwards.
Scrivorn Records the session and writes the recap, builds a page per person and place as they appear, and gives your players a portal to read between games. →Related: how to record your D&D session · is one player dominating your table?