For a puzzle, the question might be whether players understand a rule without a long explanation. For an action game, it might be whether movement and one encounter feel satisfying. For a learning activity, it might be whether the choices help people understand the intended lesson.
Keep the test small enough to change
Use temporary art and a short sequence when that is enough to test the idea. Spending heavily on content before the central interaction works can make the team reluctant to change it. The prototype should expose uncertainty while changes remain manageable.
Define what the player can do, what feedback they receive, and how the test ends. Include enough context for the intended audience, but avoid adding unrelated systems just because they appear in the long-term vision.
Decide what evidence would help
Observe a few appropriate players and record specific behavior. Where do they hesitate? What do they try? Do they understand why they succeeded or failed? Asking only whether they liked it may miss the useful detail.
After the test, separate findings from ideas. One confusing instruction may need a small change; a mechanic nobody understands may need a different approach. A prototype does not prove commercial demand or a complete production budget. It helps the team make a better-informed decision about the next playable milestone.