Types of survey questions: 12 formats, with examples and when to use each
Survey question types
✓ pick the format that matches the decision you need to make
On this page
TL;DR
Survey questions come in two families. Closed questions (single-select, multi-select, yes/no, rating scales, Likert, NPS, matrix, ranking) give you numbers you can count and compare. Open questions (short and long text) give you the reasons behind those numbers. Most good surveys use mostly closed questions, then one or two open ones where the why matters.
Choose the type by working backwards from the analysis. If you will report a percentage, you need a closed question. If you will quote people, you need an open one. If you cannot say what you would do with the answer, cut the question.
Survey question types at a glance
Type | What you get | Use it to | Watch out for |
|---|---|---|---|
| Single-select | What you getOne answer from a list | Use it toSegment people, ask about a single fact or preference | Watch out forOptions that overlap or miss a common answer |
| Multi-select | What you getAny number of answers | Use it toMap which tools, features or problems apply | Watch out forLong lists people skim; results that sum past 100% |
| Yes/no | What you getA binary split | Use it toScreen or route people with skip logic | Watch out forForcing a yes or no onto a question that has a middle |
| Rating scale | What you getA score on a numbered scale | Use it toMeasure intensity: satisfaction, ease, importance | Watch out forUnlabelled scale points that people read differently |
| Likert scale | What you getAgreement with a statement | Use it toMeasure attitudes over time with a consistent item | Watch out forStatements that ask two things at once |
| Net Promoter Score | What you get0 to 10 likelihood to recommend | Use it toTrack loyalty against a common benchmark | Watch out forTreating the score as an explanation |
| Matrix | What you getA set of items rated on one scale | Use it toCompare features or attributes side by side | Watch out forStraight-lining on long grids, mobile fatigue |
| Ranking | What you getAn order of priority | Use it toForce trade-offs between a few options | Watch out forMore than 5 to 7 items; hard on mobile |
| Short text | What you getA word or a phrase | Use it toCapture names, titles, the one word people use | Watch out forExpecting depth from a one-line box |
| Long text | What you getA written explanation | Use it toGet the reason behind a score or choice | Watch out forToo many of them; answers get shorter each time |
| Numeric | What you getA number | Use it toAsk counts and amounts (team size, studies a month) | Watch out forFree entry where a range would be easier to answer |
| Demographic | What you getWho answered | Use it toCut the results by role, company size or region | Watch out forAsking for more than you will use |
Closed survey questions
Closed questions give people a fixed set of answers. They are quick to answer, easy to analyze, and the only way to report a percentage you can defend. Their weakness is that you only learn about the options you thought to offer.
1. Single-select (multiple choice, one answer)
The respondent picks exactly one option. It is the workhorse of most surveys because the results read cleanly as shares of the total.
Example: “Which best describes your role?” Researcher · Research Ops · Designer · Product manager · Other (please specify)
Use it when the answers are mutually exclusive. Make sure the options cover the realistic range, and add “Other” or “None of these” so nobody is forced into a wrong box.
2. Multi-select (multiple choice, many answers)
The respondent ticks every option that applies. It is the right choice when more than one answer can be true at once, such as the tools a team uses or the problems they hit.
Example: “Which of these do you use to recruit research participants? Select all that apply.”
Keep the list short enough to read properly, and remember the percentages will add up to more than 100%. If you need to know which option matters most, follow up with a single-select or ranking question.
3. Yes/no (dichotomous)
Two options and nothing in between. Yes/no questions are most useful for routing: send people who answer yes down one path with skip logic and everyone else down another.
Example: “Have you run a usability test in the last 6 months?”
Avoid them for opinions. “Was the checkout easy?” forces people who found it mostly easy, with one confusing step, to pick a side, and you lose the detail you needed.
4. Rating scale
The respondent scores something on a numbered scale, usually 1 to 5 or 1 to 7. Rating scales measure intensity, which yes/no and single-select cannot.
Example: “How satisfied are you with how long it takes to recruit participants?” 1 (very dissatisfied) to 5 (very satisfied)
Label at least both ends, and ideally every point. Keep the same direction on every scale in the survey, with the low number always meaning the negative end, so people do not misread a reversed item.
5. Likert scale
A Likert item is a statement people agree or disagree with, usually on a 5 or 7-point scale from strongly disagree to strongly agree. Likert items are good for attitudes you want to track over time, because the same statement can be asked again next quarter.
Example: “Our research findings reach the people who make product decisions.” Strongly disagree · Disagree · Neither · Agree · Strongly agree
Each statement should say one thing. “Our research is fast and well-respected” asks two questions, and an answer of “agree” tells you nothing about which half they meant.
6. Net Promoter Score (NPS)
NPS asks how likely someone is to recommend you on a 0 to 10 scale, then subtracts the share of detractors (0 to 6) from the share of promoters (9 to 10). It is popular because it is one number that can be compared across teams and over time. Our NPS guide covers how to calculate and read it.
Example: “How likely are you to recommend Great Question to a colleague?” 0 (not at all likely) to 10 (extremely likely)
The score tells you that something changed, not why. Always pair it with an open question asking for the main reason behind the rating.
7. Matrix
A matrix (or grid) asks people to rate a set of items on the same scale in one table. It saves space and makes side-by-side comparison easy.
Example: “How important is each of these when choosing a research tool?” rows for recruitment, analysis, repository, security, price; columns from not important to very important
Keep grids short. On long matrices people start giving every row the same answer, and grids are awkward to complete on a phone. Five to seven rows is a sensible ceiling.
8. Ranking
Ranking asks people to put options in order. It is the strongest way to force a trade-off, because nobody can rate everything as very important.
Example: “Rank these four improvements from most to least valuable to your team.”
Ranking gets hard fast. Above five to seven items, people guess at the middle of the list. If your survey tool has no ranking question, a short series of “which matters more” single-select questions gets you most of the way.
Open survey questions
Open questions let people answer in their own words. They are slower to answer and slower to analyze, but they are where the surprises come from: the problem you did not list, the word customers use that your team does not.
9. Short text
A single line for a short answer: a job title, a tool name, the one word someone would use to describe your product.
Example: “In one word, how would you describe your current research process?”
10. Long text (open-ended)
A paragraph box for explanations. The most useful long-text question in most surveys is the follow-up to a score: “What is the main reason for your rating?”
Example: “What is the hardest part of getting research in front of decision makers?”
Use two or three at most. Each open question asks for effort, and answers get noticeably shorter as a survey goes on. Put the one that matters most early.
Questions that describe the respondent
11. Numeric
A numeric question collects a number: team size, studies run per quarter, years in role.
Example: “How many researchers are on your team?”
If people will not know the exact figure, offer ranges as a single-select question instead. “1, 2 to 5, 6 to 20, more than 20” is easier to answer than a blank box, and easier to chart.
12. Demographic
Demographic questions describe who answered: role, seniority, company size, industry, region. They are what let you cut the rest of the results by segment. Our demographic survey template has a ready-made set.
Ask only for what you will actually use to split the data, and put these questions at the end unless you need them for screening. Starting a survey with personal questions makes people more likely to drop out.
How to choose the right question type
Start from the decision, not the question. Write down what you will do differently depending on the answer, then pick the format that produces that answer.
- You need a percentage to report: single-select, multi-select or yes/no.
- You need to know how strongly people feel: a rating scale or Likert item.
- You need a trend you can track every quarter: NPS, or the same rating or Likert item asked the same way each time.
- You need priorities: ranking, or a matrix of importance ratings.
- You need the reason behind a number: one long-text question straight after it.
- You need to compare groups: demographic or numeric questions you can segment by, usually at the end.
If you are not sure what the options should be yet, run a handful of interviews or a short survey with open questions first, then turn the most common answers into a closed question. That order stops you from writing a list that leaves out what people actually say. For more on wording, see our examples of good and bad survey questions.
Common mistakes with survey question types
- Using an open question where a closed one would do. Fifty free-text answers to “What is your role?” take an hour to clean up. A single-select takes none.
- Overlapping ranges. “1 to 5” and “5 to 10” put anyone answering 5 in two buckets.
- Mixing scale directions, so 1 means good on one question and bad on the next.
- Leaving out a neutral or “not applicable” option, then reading forced answers as real opinions.
- Long matrices on mobile, which produce straight-lined answers and drop-off.
- Asking for the why on every question. Save open questions for the two or three places where the reason changes what you do.
Survey question types in Great Question
Great Question's survey builder has 13 question types: short text, long text, email, number, website, yes/no, single select, multi select, linear scale (which covers rating and NPS-style questions), date, location, matrix, and an info block for instructions or images. You can add skip logic and branching, then see a chart for every question in the results.
Two gaps are worth knowing before you build. There is no dedicated ranking question, so use a matrix or a short run of single-select questions instead. There is also no option to randomize question or answer order yet.
Because surveys run alongside your participant panel, you can send them to your own customers, filter by the attributes you already hold, and follow up with an interview when an answer needs more depth. Ready-made templates include the NPS survey, the SUS survey and the software feedback survey.
Frequently asked questions
What are the main types of survey questions?
The main types are single-select, multi-select, yes/no, rating scale, Likert scale, Net Promoter Score, matrix, ranking, short text, long text, numeric and demographic questions. The first eight are closed questions with fixed answers. Short and long text are open questions answered in the respondent's own words.
What are the five types of survey questions?
When people list five, they usually mean multiple choice, rating scale, Likert scale, ranking and open-ended questions. That covers the main shapes: choosing, scoring, agreeing, ordering and explaining. Yes/no, matrix and demographic questions are variations on those.
What is the difference between open and closed survey questions?
Closed questions give respondents a fixed set of answers, so the results can be counted and compared. Open questions let people answer in their own words, which is slower to analyze but tells you why. Most surveys work best with mostly closed questions and one or two open ones where the reason matters.
How many questions should a survey have?
As few as you can justify. Every question should map to a decision you will make. Shorter surveys get more completed responses, and open questions cost respondents the most effort, so limit those to two or three. See how to increase survey response rates for more.
Is a Likert scale the same as a rating scale?
Not quite. A Likert item measures agreement with a statement, from strongly disagree to strongly agree. A rating scale scores something directly, such as satisfaction from 1 to 5. Both produce ordinal data and are analyzed the same way.
