Every OSR game offers the referee advice on how to think about calling for a roll, but I kept noticing I was still running things like modern D&D.
I would read the advice and nod along. There are little gems of wisdom on this topic tucked into every system I've run. Intellectually, I understood them, but I struggled to turn them into a low-load heuristic I could use at the table. Under pressure, I fall back to "Roll a d20, add this bonus, let's see if you succeed."
Recently, while running Cairn, I stumbled onto a little hack that finally made it easy to run without thinking. Underneath it, I realized there was a bigger principle, but we'll get to that in a moment.
The Hack:
Don't check for success, save against consequence.
If you can't immediately explain a specific consequence of failure to the player, don't ask for a roll. You will freeze up and miss some, and you'll sometimes fall back into modern habits. That's fine. Nothing breaks. Just move on.
But when you do ask for a roll, and they fail, the consequence must happen immediately. That's the whole hack.
Let's look at it in practice:
GM: "Explain your method so I can understand how quick or noisy it might be. Are you using tools? Just kicking it in?"
Player: "I've got this crowbar, and I'm forcing it into the jamb around the lock to break the door frame where the deadbolt slots in."
GM: "That's not as noisy as kicking it down, but it might take some time. Make a roll to see if you get it open before a guard patrols by. If you fail, we will have an encounter."
The player explains their plan, giving us time and context for an obvious consequence to present itself. We state a specific consequence that will happen on failure, and we justify it in the fiction. The player isn't forced into anything; this is negotiation. They can go for it, back off, or suggest ways to mitigate the consequences.
We completely skip the question "Do you open the door?" in favor of two more interesting events. First, they must ask, "Do I actually want to take this risk?" and then, assuming they do, we find out, "Do I get caught?" These are both more compelling experiences for everyone at the table than "how many times does Bob have to roll the dice to get through the door?"
I said earlier, I discovered a big principle hiding in there. Here it is
Never decouple Failure from Consequence.
It's caused every bad thing that's ever happened to me when calling for a roll, and the "save against consequence" hack sidesteps it because if there is no consequence, there can't be a roll.
Never decouple Failure from Consequence.
Let's unpack a few common failure modes.
"Nothing happens" isn't a consequence.
Any roll whose possible outcomes are "It happens" or "Nothing happens" is a waste of time, because players will just keep rolling until it works. It's worse than a waste of time: you signaled that there was a consequence when you asked for the roll, then demonstrated that there wasn't.
This is how dog-piles happen. This is how the wizard with 6 Strength pops a door that the fighter with 18 Strength couldn't manage.
If the fighter fails the roll to jimmy open the door, then either the door is completely barred from the other side, or they made too much noise and attracted a monster, or they must spend a resource, and so on. If they roll at all, the result is final, and the consequence happens. There can't be a dog-pile if the party is now busy in combat.
Random encounter checks aren't a consequence.
There is no real tension when a failed roll just leads to another roll. It's too abstract, especially when that second roll has a 5-in-6 chance of resulting in "nothing happens."
Instead, tell the player they are making a save to see whether a patrolling guard arrives before they finish with the door. Now the whole table leans in as the fighter rolls, because something concrete is on the line.
Players metagame failed checks to avoid consequences.
Failed stealth, perception, and "find traps" checks tend to get squishy. The player now knows a hidden consequence might be floating around, and they still have action economy to try to avoid it. That happens because the consequence wasn't clearly coupled to the roll.
Don't check to see if they hide; check to see if they get caught. Don't roll to see if they "spot the trap"; roll to see whether they blunder into it before they realize something is wrong. When you phrase rolls as saves against clear consequences, you close off this gray zone.