One of the fastest ways to break immersion is giving the player correct information from the wrong source. The problem is not the information itself. The problem is that the world stops feeling real when somebody knows something they should not know. A random villager should not know the command officer’s personal safe code unless there is a believable reason. A janitor should not know the war plan. A prisoner should not know the exact layout of a classified bunker unless they were held there long enough to learn it. Information should feel like it came from the world, not from the GM pushing the story forward.
The easiest rule is this: the source must match the information.
Before you give any clue, answer three questions.
This is the most important question.
Did they see it?
Did they hear it?
Did they read it?
Did they work with it?
Did someone tell them?
Did they guess correctly from experience?
Did they survive something related to it?
A person can only give information that fits their access. A cook may know when officers eat, which doors are used most, who skips meals, and which guards sneak extra rations. That same cook probably does not know the contents of a sealed military file. A clerk may know names, schedules, transfers, passwords written down in the wrong place, or which officer signs what. A mechanic may know blind spots in maintenance tunnels, generator weaknesses, and vehicle routes. A servant may know habits, moods, routines, and private embarrassments. Each source should give the kind of truth their life puts them near.
Even if someone knows something, that does not mean they will just hand it over.
Are they scared?
Are they angry?
Do they want revenge?
Do they trust the player?
Are they trying to manipulate the player?
Do they think the information is harmless?
Do they want payment, rescue, protection, status, or mercy?
Good information comes with motive. People do not become clue dispensers just because the player asked nicely. A prisoner might talk because they want out. A clerk might leak because they hate their superior. A guard might reveal something while bragging drunk. A grieving widow might share details because she wants justice. When motive is clear, the clue feels earned.
This is where a lot of games fail. Not all knowledge comes in perfect, clean sentences.
Real information often comes as:
partial memory
rumor
routine knowledge
technical jargon
wrong assumptions
outdated facts
fragments overheard
documents missing context
visual details the player must interpret
A civilian might say, “The officer always goes upstairs after evening meal and sends everyone away.” That is believable. Saying, “His personal lockbox code is 4-7-2-9 and the ledger inside proves the conspiracy,” is not, unless there is a strong reason.
Information should follow position.
Low-level people know local truth. They know habits, gossip, schedules, moods, shortcuts, and visible routines.
Mid-level people know procedures. They know chains of command, standard responses, routes, regulations, supply timing, and common security habits.
High-level people know plans. They know intent, strategy, hidden tensions, real objectives, and protected secrets.
Documents know specifics. Logs, letters, maps, orders, codes, manifests, terminals, journals, and reports are where exact information should often live.
This is the easiest fix for unbelievable exposition. Put exact info in the hands of records or people whose jobs require it. Put human, usable, street-level info in the mouths of ordinary people.
People know what they are near.
A housekeeper may know the officer drinks alone, locks one drawer, keeps a spare key in his boot room, and dismisses staff before opening the wall safe. That is excellent information because it comes from proximity.
That same housekeeper probably does not know the safe combination itself. But maybe she once saw the officer enter it while polishing a mirror. Maybe she noticed he always taps the same birthday into the dial. Maybe she found a torn note in the trash. Now it works, because the clue came from what she could reasonably observe.
The question is never “Can this NPC help the players?”
The better question is “What would this NPC realistically notice?”
Different sources should give different kinds of clues.
Good for rumors, habits, schedules, visible events, strange behavior, who comes and goes, what feels wrong, and which places are feared.
Good for routines, access points, maintenance routes, storage areas, daily timing, staff personalities, and small vulnerabilities.
Good for patrol paths, checkpoints, standard orders, restricted areas, passwords that rotate, unit behavior, and what has recently changed.
Good for names, paperwork, transfers, logs, room assignments, codes written where they should not be, signatures, supply records, and bureaucratic weak points.
Good for intent, real stakes, cover-ups, classified areas, layered security, emergency procedure, and the true reason something matters.
Best for exact details: maps, codes, manifests, account books, names, dates, proof, and technical facts.
This keeps information feeling grounded. The wrong source creates convenience. The right source creates immersion.
The player should usually not get the full truth in one clean drop. Better information comes in steps.
First they hear the officer dismisses everyone before going into his office.
Then a servant mentions he never trusts the same guard twice.
Then a clerk says a locked black ledger is always on his desk after inspection days.
Then a note or terminal reveals the safe exists.
Then the code is found in a believable place: hidden in a prayer book, disguised in an account sheet, memorized by a blackmailed aide, or learned from watching him use it.
Now the information feels discovered instead of gifted.
The more precise the clue, the more justified its source should be.
General info is cheap. “There are more guards at night” can come from almost anyone nearby.
Specific info is expensive. “The west service hatch opens for seven minutes after the 0200 pressure purge” should come from a maintenance worker, shift supervisor, technical log, or direct observation.
Perfect info is most expensive. “The command officer’s personal lockbox code is 8-1-4-2” should come from one of only a few believable places: the officer himself, someone who watched him, a trusted aide, a written record, a hidden backup key, or a security flaw the players uncover.
Exactness requires justification.
A believable world is full of flawed information.
A drunk guard remembers the wrong patrol time.
A villager repeats rumor as fact.
A bitter ex-servant exaggerates.
A prisoner lies to get moved.
A clerk has outdated paperwork.
This does not mean you should trick the player constantly. It means information should feel human. Some clues are solid. Some are useful but incomplete. Some are biased. The player learns to judge sources, not just collect answers.
Good information should open the story, not flatten it.
A servant tells the player the officer hides private papers in his office. Good.
But she also says he changed his habits after a recent theft. Now the clue helps, but the problem stays alive.
A mechanic reveals a maintenance tunnel. Good.
But the tunnel floods during generator bleed-off. Now the clue is real, useful, and still dramatic.
This keeps information from feeling like a cheat code.
An old woman in town says, “The officer’s lockbox code is 2-9-1-6, and the payroll ledger inside proves he is funding the enemy.”
This feels fake unless there is a huge explanation.
An old laundress says, “I never saw the inside of his safe, but every seventh day he sent everyone off before opening the painting in his office. Always after payroll came in. He’d mutter his daughter’s birthday under his breath when he used the dial. I remember because he only ever sounded soft when he said it.”
Now the source makes sense. She knows habits, timing, and a possible code basis. The player still has to use it.
Before giving the player information, ask:
Access — Could this source actually know this?
Motive — Why are they giving it up?
Form — Would they know it exactly, or only partially?
Risk — Does sharing it put them in danger? If so, why do it anyway?
Usefulness — Does this clue help without solving everything by itself?
If the clue passes those five checks, it will usually feel believable.
Do not ask, “How do I get the clue to the player?”
Ask, “Who in this world would naturally carry this piece of truth?”
That one change fixes almost everything.
When information comes from the right place, the world feels real. The player stops feeling like the plot is feeding them answers and starts feeling like they are uncovering them. That is what you want: not random exposition, but knowledge with weight, origin, and logic.