It’s 10:52. You said you’d wrap at eleven. The party has just kicked in the door to the counting house, the guards are rolling initiative, and your cleric has that look that means she is about to do something expensive with a spell slot. Somebody’s partner has walked past the webcam twice. The bard is on his phone checking what time his train is. And you, the person who is supposed to be steering this thing, are quietly doing the math on whether a six-guard fight can possibly finish in eight minutes, knowing full well it cannot, and knowing that if you push through, the last hour of the session will be played by people who have already left.

Everybody has advice about how to start a session. Almost nobody talks about how to end one, and I’ve come to think the ending is the more important of the two. How you end tonight decides how you start next week, whether next week gets scheduled at all, and how much of tonight your table will still remember when it does. It is a five-minute piece of craft and it is worth doing on purpose.

The last fifteen minutes are for landing, so stop taking off

The single most useful habit I’ve built as a DM is a glance at the clock at T minus fifteen. From that moment on, nothing new comes into the session. No new room. No new NPC with a name I’ll have to remember. No fight, unless the fight is the ending. Everything after that glance is about landing what’s already in the air.

That means at 10:45 the counting house door stays shut, and if the party wants to kick it in I let them get all the way to the doorway, and I let them hear the guards on the other side, and I stop there. What was going to be a fight I couldn’t finish becomes a picture nobody will forget: five people, one door, a decision.

The alternative, the fight you start at 10:52, ends one of two ways. Either it runs forty minutes over, and the people with trains and partners and kids play the back half badly and feel guilty about it, or you call it mid-round and spend the first twenty minutes of next session reconstructing initiative order, hit points, and who was where. Both are worse than the door.

Ending on time is a campaign survival skill

I wrote a whole post about session zero that argued most campaign deaths are scheduling deaths, and the ending is where scheduling actually gets decided. Every session that runs long teaches the table that “we play until eleven” is a suggestion. The player who has to leave on time starts feeling like the person who ruins the fun, so they stop saying they have to leave, and then they stop coming to the sessions they’d have to leave early from, and six weeks later you’re trying to find a night that works for four people instead of five.

The DM who ends at eleven when they said eleven, even in the middle of something great, especially in the middle of something great, is the DM whose table is still meeting a year later. Running over feels generous. It reads, to the person watching the clock, as a small betrayal.

Cliffhangers are overrated. End on a choice.

The standard advice for a session ending is the cliffhanger: the villain steps out of the shadows, the floor gives way, roll credits. I’ve done a lot of these and I’ve mostly stopped, for a reason that took me a while to see. A cliffhanger ends the session on my move. The players spend the week waiting to see what I do next.

Ending on a choice puts the week in their hands instead. Give them the door and the sound of boots and the fact that the vault is somewhere in this building, and let them start arguing about whether to go loud or quiet, and when it’s a real argument, when two players genuinely disagree, that’s when you say “and that’s where we’ll pick up.” Now they think about it in the shower. They text each other on Thursday. The one who wanted to go quiet shows up next week with a plan, and the first ten minutes of the session run themselves, because your players spent the week doing your prep for you.

You still get to have your reveals. Put them at T minus thirty and let the choice they create be the ending. The villain in the doorway is a great beat. “What do you do about the villain in the doorway, and we find out next week” is a great ending.

Give the fight its turn, then stop before the bill

If the ending has to fall in combat, and sometimes it does, stop right after the round where the fight turns. The paladin lands the smite that takes the captain to his knees. The wizard’s wall goes up and half the guards are on the wrong side of it. That is the moment. Call it there, before anybody has to deal with what the turn costs.

Two reasons. The obvious one is emotion: your players leave feeling like they’re winning, and they’ll come back to find out if they’re right. The less obvious one is memory. Ending in the middle of a fight means somebody has to write down initiative order, everyone’s hit points, every condition, and where the tokens are, and it will be reconstructed wrong next week no matter how carefully it’s written. Ending right after a big turn means the fight state is simple. “Captain’s down to a sliver, wall’s up, three guards behind it, you’re up first.” Ten seconds to restore, and everybody remembers, because it was the last thing that happened.

The five-minute wrap that earns you next week

Once the story stops, don’t let the call end. The five minutes after “and that’s where we’ll pick up” are the highest-leverage five minutes of the night, and most tables spend them saying goodbye.

Book the next session before anyone leaves. Do this first, while everyone is still on the call and still in a good mood from the door. “Same time next Tuesday?” gets a yes in eight seconds from five people who are right there. The same question in the group chat on Thursday gets two thumbs up, one “probably,” and a person who doesn’t see it until Saturday. Scheduling from the table is close to free. Scheduling from a group chat is a part-time job.

Ask each player one sentence. “What is your character thinking right now?” Go around the table. This takes two minutes, it’s the best cool-down there is, and you learn things you’d otherwise never hear: that the rogue has quietly decided the cleric is going to betray them, that the fighter is only here for the money and it’s starting to bother him. Half of my best sessions were built out of an answer someone gave in that circle.

Name the moment. Say out loud the one thing from tonight that will still be funny or terrible in a month. “The bit where Ansel tried to bribe the guard with the guard’s own coin purse. That’s going in the recap.” Naming it makes it canon, it gives the player who did it a little glow to carry out the door, and it tells you what tonight was actually about, which is rarely what you prepped.

Write two lines for future you. Where we stopped, and what’s about to happen. “Outside the counting house, door open, guards inside. Party arguing loud vs. quiet; Mira wants quiet.” That’s it. Do it before you close the laptop, because by tomorrow it’s gone. I’ve argued elsewhere that prep starts by reading last session, and these two lines are what makes that reading take ten seconds instead of ten minutes.

Where the good endings go to die

You do all of that. You end on the door, on time, on a choice. You book Tuesday, you go around the circle, you write your two lines. Then Tuesday comes and you open the two lines and it says “guards inside, Mira wants quiet,” and you cannot for the life of you remember why Mira wanted quiet, or what the fighter said in the circle that made everyone laugh, or the exact thing the rogue admitted about the cleric that you swore you’d use.

The ending you engineered so carefully lives in the memories of six people, and those memories are all lightly wrong in six different directions. Somebody at the table on Tuesday will be sure the guards were already alerted. Somebody else will have forgotten the vault entirely. The choice you left them, the one that was supposed to start the session by itself, starts instead with five minutes of “wait, was the door open?”

The ending is on the recording

Loracle sits in your voice channel and records the session, then writes the recap and keeps track of the threads still open, so "where we stopped and what's about to happen" is already written down when Tuesday comes, in the words your table actually said. The two lines you scribbled at 11:02 get to be a bonus instead of the only record.

Keep the ending →

The whole thing on an index card

At T minus fifteen, stop introducing anything new. End on time even when it hurts, especially when it hurts. Land on a choice the players are still arguing about, and save the reveals for half an hour earlier so the choice can be the ending. If you must stop in a fight, stop right after the turn that changed it. Then, before anyone leaves the call: book the next session, ask each player what their character is thinking, name the moment, and write two lines for future you.

Back at 10:52. The door is open. The guards are shouting. Mira is saying “we go quiet or we don’t go,” and Ansel is already reaching for the oil flask, and you say “and that is exactly where we pick up next Tuesday. Same time?” Five yeses. Then the circle. Then the two lines. Then you close the laptop at 11:04 and the session, somehow, is still going, in five other heads, all week.