How to run a good playtest
Watch first, ask second. Players forget what confused them by the time they fill in a form, so sit quietly next to them or record the screen, note where they get stuck, and only help when they are completely lost. Then hand them the survey while the game is still fresh. Five testers find most of the big problems. Test again after you fix them.
Asking better questions
Ask about what happened, not what they would like. "Where did you get stuck?" gives you facts. "Would you like more levels?" gets a polite yes. Keep scales the same direction, low is bad and high is good, so averages can be compared between tests. Put the open questions about the best and worst moment near the end, when players have something to say. Do not ask for personal details you do not need.
What to ask at each stage
A prototype test answers one question: is the core loop fun? Ignore art and bugs. An alpha adds the tutorial, controls and readability. A beta adds performance, bugs, art and spending. A demo before launch, such as for Steam Next Fest, is about whether players would wishlist and buy, and at what price. The tool picks these for you when you change the stage.
Reading the results
The recommend score, or NPS, is the share of players who answer 9 or 10 minus the share who answer 0 to 6. Above 0 is fine for an early test and above 30 is strong. Averages on 1 to 5 scales are most useful compared between tests of the same game. The common words and quotes show what players noticed, which often matters more than the numbers.
Is anything uploaded?
No. The survey and the responses CSV stay in your browser.