The most useful skill we have picked up as a studio is not level design, or shader tricks, or writing tight little bits of code at midnight. It is learning to sit on our hands during a playtest and say absolutely nothing. It is the hardest thing in the building, and it is where almost every good decision we have made actually came from.
The hardest part is keeping your mouth shut
When you hand someone your game, every instinct screams at you to help. They pause, and you want to say “oh, the button is in the corner”. They frown, and you want to explain the clever thing they have not noticed yet. Do not. The moment you jump in, you have stopped learning anything, because now you are testing a game that comes with a developer attached. Nobody at home gets that version.
So we set a rule for ourselves before the player even sits down: we are allowed to say hello, and we are allowed to say thank you. In between, we mostly shut up and watch.
Watch their hands, not their face
Faces lie, politely. People will smile and nod and tell you it is great because they are lovely and they do not want to hurt your feelings. Hands do not lie. Where does the cursor hover when they are unsure? Which button do they mash when they panic? Where do they stop, backtrack, and try the same wrong thing twice? That is the real feedback, and the player is giving it to you without saying a word.
We take notes on behaviour, not opinions. “Said it was fun” goes in a small column. “Tried to leave through the wall three times” goes in a very large one.
The first sixty seconds tell you everything
If we could only watch one minute of a playtest, it would be the first. That is where a player decides whether your game makes sense, whether it respects their time, and whether they want to keep going. Confusion in the opening seconds is not a small problem you can patch with a tooltip later. It is the whole first impression, and you only get one.
So we watch that first minute like hawks. Did they understand what to do without being told? Did the game teach itself? If the answer is no, that is the next thing we fix, before anything shinier further in.
Ask what they did, not what they want
When the test is over and we are finally allowed to talk, we are careful about the questions. Players are brilliant at spotting problems and surprisingly unreliable at designing solutions. “I got lost here” is gold. “You should add a minimap” is a guess, and often the real fix is something else entirely.
So we dig into what happened, not what they would build. Where did you expect that to take you? What were you trying to do right there? The answers point at the actual problem, and the actual problem is almost never the one the player names first.
One fresh pair of eyes beats a room of friends
Friends and family are wonderful and we love them, but they are the worst possible testers. They want you to win, they already half know how your game works, and they will forgive things a stranger never would. One person who has never seen the game, given a controller and left alone, will teach you more in ten minutes than a week of encouraging nods from people who care about you.
It stings a little, which is how you know it is working. The goal of a playtest is not to feel good. It is to find the stuff that is broken while it is still cheap to fix.
So the next time you run a test, try the Sharkbowl approach: hand it over, zip it, and watch the hands. Write down what people do, not what they say, and treat that first awkward minute as the most honest review your game will ever get. (Finn insists he gives excellent feedback. Finn is a mascot and has never finished a level. We still watch his hands.) No sharks were harmed in the making of this post, though a few egos were gently bruised, as intended.
