Playtesting
Learning paths through this topic
Did the jump scare really land? Read it from how the tester stopped moving
Testers rarely say "that scared me", but their hands do. This entry shows how to tell a scare that landed from one that was shrugged off, without a camera and without asking, and why a fixed pause length misjudges calm and fast players alike.
After how many scares do testers stop being scared?
The tenth scare never hits like the first. This entry finds the moment where your testers get used to it, and explains why the obvious average over everyone gives a rosier picture than the truth, because the easily frightened testers leave early.
Testers played with the sound off: how many scares did they really miss?
In a scary game, half of the fear lives in the audio. If some testers had the speakers muted, your landing rates are mixed up with a setting nobody meant to test. This entry separates the two and says which scares really need sound.
Scares planned a minute apart arrived ten seconds apart for one tester
If two scares come too close together, the second one is wasted and the player just feels hassled. This entry shows how to measure the real gap each tester got, and why the clock on the wall gives a false picture the moment someone opens a menu.
Some testers played twice: keep only the run where they didn't know the story
A tester who replays knows every twist and clicks through at speed, which makes your story look faster and less scary than it is for new players. This entry keeps the right runs, including the ones interrupted by a crash that were not really replays.
A tester solved my puzzle, but did they understand it or stumble into it?
A finished puzzle in a playtest proves less than it seems, because luck and trial and error finish puzzles too. This entry shows how to put a twin of the puzzle right behind it and turn the second result into an honest share of testers who got the idea.
Five testers, thirty levels: who plays which so no level is left out?
Left to themselves testers play the first levels again and again and never reach the last ones. This entry shows how to hand out levels so that every level gets the number of looks you want, new or changed levels get more, and nobody plays the same one twice.
A short quiz after the story shows which facts nobody remembered
If testers cannot say who the stranger was or why the door was locked, the plot has a hole in the player's head, not on paper. This entry turns a handful of quiz answers into a fair score per fact, correcting for people who simply guessed.
Which combinations of story choices has no tester ever had at the finale?
A finale that reacts to many earlier choices has far too many combinations to test all of them. This entry checks the much smaller set of pairs of choices, lists the pairs no tester has met, and ignores pairs the story makes impossible.
Before testers arrive: can every scene be reached, and can every ending be won?
Stories with keys, trust and other remembered facts hide dead ends that only appear when you play a very particular way. This entry walks the story map for you and names the rooms nobody can enter, the doors that need a thing nobody can get, and the loops without exit.
Which branches of my story has no tester walked yet, and which routes should they try?
With a few testers you cannot see everything, and asking them to "try other things" gets you the same favourite route. This entry lists the untouched branches and then hands out the smallest set of routes that would cover them all.
Some sessions look inhuman: how to find automatic tapping in my numbers
In idle and tapping games a few sessions play with a rhythm no hand can keep, and they bend every average of progress and speed. This entry shows the one property that gives a clicker script away, what limit to use and when not to judge at all.