The playtests you most need are the ones the moderator stopped moderating.
What most studios do
Sit two people next to a stranger, hand them a build, and ask "so, what did you think?" ninety seconds in. Every studio has run this playtest. Almost nothing changes as a result.
The shape that works
Three parts, in order, under 45 minutes:
- A falsifiable hypothesis. Not "is it fun." "The player discovers the parry within 90 seconds without a tutorial pop-up."
- Silent observation. The moderator stays quiet for the first five minutes. Log pauses, retries, laughs, curses. Do not rescue.
- Behavior before opinion. Debrief opens with "walk me through the last five minutes." Only later: "if you could delete one thing from this build, what?"
The decision line
Every report ends in one of three words: ship, iterate, cut. If your report ends in a summary, it wasn't a playtest, it was a demo.
The tell
If the team leaves the debrief and nobody has changed their mind about anything, the playtest failed, not the build.
Draft yours with the Playtest Script tool.