How to write playtest questions that give you decisions (with a ready-to-use form template)

A playtest gives you two things: a build in someone else's hands and a form to collect what they learned. Most studios put all the care into the first part and treat the second as a chore. That is where the value leaks out — a form full of vague, leading or overlapping questions turns a good session into a pile of opinions nobody can act on.

This guide shows how to write questions that produce decisions, and ends with a form template you can copy, adapt and reuse in your next test.

Start from the decision, not the question

Before writing a single question, list the decisions this playtest has to support. Do we keep the second boss or cut it? Is the tutorial too long? Should the crafting menu open from the pause screen? Then write one question per decision — and drop every question that serves none.

A question earns its place when the answer changes something. If every possible answer leaves you doing the same thing, it is curiosity, not research, and it costs testers the time you asked for.

Use the six question types well

Each type does a different job. Use the smallest one that still answers what you need.

Short text — one or two lines: a label, a word, a quick "what did you expect here?". Long text — the stories you cannot predict: what surprised you, what made you stop. Rating (1–5) — anything you want to compare across versions or testers: clarity, difficulty, fun, pacing. Single choice — mutually exclusive options you want counted cleanly: which part felt longest. Multiple choice — several answers can be true at once: which systems you would keep. Yes / no — facts and gates: did you find the save option, did you reach the ending.

Rating and choice questions are cheap to answer and easy to compare. Text questions are expensive for the tester and slow to analyse, so keep them for what only words can capture.

Rules that keep a form honest

Ask one thing per question. "Was the combat fair and fun?" hides two answers in one field.

Avoid leading phrasing. "How much did you enjoy the new combat?" assumes enjoyment. "Describe the new combat" does not.

Ask about behaviour, not intention. "Would you buy this?" is a guess. "Did you replay the level?" is a fact.

Keep the scale consistent. If 1 means "very bad" in one question and "very easy" in the next, you will misread your own results.

Say what the ends of a scale mean. "Rate the difficulty from 1 to 5 (1 = too easy, 5 = unfair)" beats a bare number.

Keep it short. A dozen well-written questions answered by ten people beat thirty answered by four. Response rate is a real number in your results, and a long form is the fastest way to damage it.

Put the easy questions first, and place the open text question you care most about while attention is still high — usually near the middle, not at the very end.

Mistakes that waste a playtest

The general impression. "What did you think?" produces polite silence or an essay. Ask what happened and how it felt.

The future promise. "Would you play this again?" is answered by nice people who mean well. Ask whether they kept playing after the optional break.

The invisible question. A question that assumes context the tester never had — "How balanced is the late game?" in a twenty-minute test.

The unlabelled scale. A rating with no anchors, where half the room reads 5 as "hard" and the other half as "excellent".

The duplicate. Two questions asking the same thing in different words double your analysis time and add nothing.

The form nobody reads back. If you do not look at the answers in the same week, the playtest loses most of its value.

A ready-to-use feedback form template

Copy this, rewrite the words in brackets for your game, and cut anything that does not serve a decision. The type in parentheses is the question type to use.

— Before you play 1. Which genres do you usually play? (Multiple choice) 2. How often do you play games like [your genre]? (Single choice) 3. What did you expect this game to be about before playing? (Short text)

— Right after the session 4. How clear was the objective in the first five minutes? (Rating 1–5) 5. How would you rate the difficulty of this section? (Rating 1–5) 6. How was the pace — too slow, too fast, or about right? (Single choice) 7. Which part felt the longest? (Single choice) 8. Which elements would you keep if the rest had to be cut? (Multiple choice) 9. Did you ever feel stuck without knowing why? (Yes / no) 10. Did you find how to [key action: save, change difficulty, leave a level]? (Yes / no) 11. Describe the moment you felt most frustrated. (Long text) 12. Describe the moment you enjoyed most. (Long text) 13. What did you expect to happen that didn't? (Short text) 14. Anything else we should know? (Long text)

How to adapt the template

For a prototype: keep the clarity and frustration questions, drop the polish ones, and add "What did you expect to happen when you tried [core action]?".

For a vertical slice: add difficulty and pacing ratings, and ask which part felt longest so you can see where attention drops.

For a balancing pass: add one rating per weapon, enemy or boss, and a single-choice question about which felt weakest.

For a localisation check: add a short text question about wording that felt strange, and a yes/no for each screen you are unsure about.

Reading the results

Look at the numbers first: averages, the spread of answers, and how many people finished the form. A rating of 3.4 where most answers are 2 and 5 is a different problem from a 3.4 where everyone answered 3. Then read the text answers and mark the comments that repeat — repetition is the signal, not the loudest single opinion. Group everything into fix now, investigate, and keep as is, then tell your testers what changed because of them.

How Playtest Hub helps

Playtest Hub keeps the form and the analysis in the same place. When you create a playtest, you write a feedback form with the six question types above — short text, long text, rating, single choice, multiple choice and yes / no — and choose which ones are required. Testers answer inside the playtest, and when it ends the Results page shows, for that playtest only: how many people participated, how many submitted feedback, the response rate, and a breakdown per question — average and median with the distribution for ratings, counts and percentages for choices, and every written answer for text questions.

That structure is why the advice in this guide matters: a question written to be counted can be summarised automatically, and a summary a team can read in ten minutes is the difference between feedback and a decision.

A good question is a small design problem: it has to be honest, answerable and useful. Write it from the decision you need, pick the smallest type that captures the answer, keep the form short, and read the results while the session is still fresh. Do that and ten testers will tell you more than a hundred opinions ever could.