Question type is one of those details that looks small until you watch people answer a survey.
The same question can feel effortless or annoying depending on the input. Ask someone to type when they only need to choose, and you add friction. Force nuance into a dropdown, and you lose the insight. Since drop-off climbs steeply once a survey feels like work, the right format is not cosmetic, it is a completion-rate lever.
Modern builders (Formaly included) support 30+ types, but ten of them do almost all the work. Here is when each one earns its place.
1. Multiple choice
Use it when the options are known and only one answer applies. This is the workhorse: fast for respondents, clean for analysis. Keep the list tight and mutually exclusive, and add an "Other" escape hatch only when you genuinely expect one.
2. Checkbox (select multiple)
Use it when more than one thing can be true. The trap is length, if respondents must scan 20 options, the question becomes a chore. Group long lists, or split them.
3. Rating scale
Useful for satisfaction, effort, confidence, or likelihood. The key is labeling the endpoints clearly: a bare 1–10 means little if nobody knows what 1 and 10 represent. Match the dimension to the question, "easy/hard" for product effort, not generic "good/bad."
4. NPS (0–10 likelihood to recommend)
A standout because it is a recognized standard, respondents already know how to answer, and you can benchmark over time. Always pair it with a "what's the main reason for your score?" follow-up, or you get a number with no why.
5. Slider
Works when the answer is continuous: budget range, intensity, frequency, importance. Do not use a slider just because it looks interactive, use it when the interaction matches the underlying scale.
6. Ranking
For tradeoffs. "Which feature matters most?" is far more useful as a ranking than as five questions where everything gets rated "very important." Keep the list short, ranking ten items is genuinely painful.
7. Matrix
Efficient when several items share one scale. Also the most dangerous type on mobile, where completion is already ~10% more fragile. Use sparingly, keep grids small, and consider splitting into individual questions on small screens.
8. Yes/No
Best for routing. "Have you used the new dashboard?" → if yes, ask about it; if no, skip. Cheap to answer, and the backbone of good branching logic.
9. Image choice
Ideal for visual preferences: concept tests, packaging, ad variants, UI options. Far better than asking someone to describe a visual comparison in words, and usually faster to answer, too.
10. Contact / open text (use with restraint)
Open text is where the real language lives (reasons, objections, stories you did not anticipate), but it is the most expensive type, so earn it by placing it after a structured answer. Contact fields (email, phone) add the most friction of all: ask only when you truly need them, explain why, and let the survey stay anonymous when it can.
The rule underneath all ten
The best question type makes answering feel obvious.
That is the standard. If the respondent has to think about how to answer rather than what to answer, the format is creating unnecessary friction, and friction is what you pay for in abandoned responses.
Formaly treats question types as part of survey quality, not just UI variety: prompt-generated surveys pick types based on what you are actually asking, so a prioritization question becomes a ranking and a nuanced objection gets open text instead of a dropdown.