Playtesting a new game properly before you sell a single ticket

Most new rooms get tested by the people who built them, plus a few friends who came on a Tuesday and said it was great. That is not a playtest. It is a reassurance ritual, and it is why rooms open with one puzzle nobody outside the building can solve.

A playtest is data collection. The tester you actually need is a stranger of roughly average ability who owes you nothing, has never met you, and has no social reason to soften the verdict.

The rest of this is how to get that, and what to do with what it gives you.

Three waves, three different questions

Testing falls apart when one session is asked to answer everything at once. Split it into waves and give each one a single job.

• Wave one: does the chain work. You and whoever built it, walking the sequence with the answers in hand. You are checking that every mechanism opens, every prop is where the puzzle assumes it is, and no step needs information the players have not been given yet. Nobody enjoys wave one. That is not what it is for.
• Wave two: is it fun. Friends, regulars, staff from the shop down the road. They know you, so they will not be honest about fairness. They will still tell you where the energy dies, and which moment they described to somebody else afterwards.
• Wave three: is it fair. Strangers. This is the only wave that tells you whether the game works for the people who will pay for it. If you only have room for one proper round of testing, make it this one.

Wave three is the one operators skip, because it is awkward to organise and likely to produce bad news two weeks before opening. Which is precisely why it earns its place.

Recruiting testers who will not flatter you

The enthusiast community will volunteer instantly, which is both the appeal and the problem. People who have played sixty rooms will solve a puzzle your actual customers cannot, and then tell you it was fine. Test with them by all means, but not only with them.

The tester you want has played one to three rooms in their life. Find them through your own mailing list, through the staff of nearby businesses, through friends of friends you will not be seeing socially next week. Comp the session, say plainly that the price is blunt feedback, and ask for it in a form people can actually deliver: tell me what you would have said about this in the car on the way home.

Test at both ends of your group size. A puzzle designed for four behaves differently with six people who cannot all reach it, and differently again with a couple who have nobody to bounce a bad idea off.

The designer says nothing and writes everything down

The most common way to waste a test is to help. Every time you clarify an instruction, confirm a hunch or laugh at the right moment, you have destroyed that datapoint permanently. If a group is stuck for fifteen minutes on something you thought was obvious, that fifteen minutes is the finding. Sit on your hands.

A designer who cannot stay quiet should not be in the room. Put them on the camera feed with a clock and a notebook.

What to record, timestamped:

1. Time in and time out of every puzzle, not just the total.
2. The point where the group split, and what the smaller half did with itself.
3. Every wrong answer offered out loud. These are the most valuable lines in your notes. A wrong answer that three separate groups reach independently is not a wrong answer, it is a second correct answer you failed to accept, or a signposting failure you have not spotted yet.
4. Every hint request, and the question underneath it. We are stuck and we do not know what this device is for are different problems with different fixes.
5. The moment somebody stopped looking at the puzzle and looked at their watch.
6. Anything they picked up that was not part of the game.

Then three questions at the door, before they have talked themselves into being polite: what was your favourite moment, where did you feel stupid rather than stuck, and what did you think you were supposed to be doing at minute forty.

Fast and bored versus slow and happy

Completion time is the number everybody records and the number that means least on its own. A group that finishes with twenty minutes to spare and files out quietly has told you the game is thin. A group that runs out of clock and then stands in your lobby arguing about the last puzzle has told you the game is close to right and the ending needs work.

So pair every time with the mood at the door. Fast and delighted is a good game with a short tail. Slow and delighted is a good game with a hint policy problem. Fast and flat is the dangerous one, because the number looks healthy on the spreadsheet.

Resist setting yourself a target failure rate before you have opened. What you are tuning towards is a finish that feels earned, and you can hear that in the room without converting it to a percentage.

Change one thing at a time

After a rough test the urge is to fix all six problems before the next group arrives. Do not. The next group then has an entirely different experience, and you cannot tell which change did what, or whether one of your fixes opened a new dead end.

Change the single thing that did the most damage, run it again, and keep a dated log of what changed after which session. That log is quietly valuable a year later, when you are re-theming and cannot remember why one clue is worded so strangely.

How many tests is enough? Until two consecutive groups of strangers get through with roughly the hint pattern you intended, and nobody hits the same wall twice. In practice that is more sessions than you planned and fewer than you feared.

Test in the conditions you will actually sell

Rooms get tested at two on a quiet weekday with the owner present and the building empty. They get sold on a Saturday night with a birthday group in the lobby, one game master covering two rooms, and a team that arrived twelve minutes late and slightly drunk. Run that version at least once before you open.

This matters more the less control you have over the space. City Sleuth partners build outdoor routes through public streets, so a test has to happen in the rain, at dusk, and on the day the square is full of a market that was not there last week. A puzzle that only works in good weather and light foot traffic is not a puzzle, it is a fair-weather liability.

Next step: keep reading the logs after you open

Launch is not the end of testing, it is the start of a far bigger sample. For the first month, read the hint logs weekly and look for the puzzle that pulls a hint from almost every group. One of those is usually a genuine break rather than a hard puzzle, and it will be visible in the logs long before it appears in a review.

Pick the next game on your build list and write down today who your wave three testers will be. If that list is empty, that is the first thing to fix. If you are curious how the same discipline works when the game board is a whole downtown rather than a room, the rest of the City Sleuth partner guide covers the outdoor version.