Most guides to quantitative questions organise them by format: Likert scales here, multiple choice there, rating scales after that.
- Start from the decision you need to make; design questions to produce that number, then choose formats accordingly.
- Make questions truly quantitative: include a unit and timeframe, use mutually exclusive exhaustive answers, and avoid leading wording.
- Design surveys for quality: screen respondents, keep surveys short with skip logic, pilot first, and prevent framing effects from question order.
That is backwards for anyone actually building a study. You do not start a research project thinking “I need a Likert scale.” You start thinking “I need to size this market” or “I need to know what people will pay.” The format is a consequence of the objective, not the other way round.
These 21 examples are organised the way research actually gets designed — by what you are trying to find out.
What Makes a Question Quantitative
A quantitative question produces a response you can count, compare across respondents, and analyse statistically. That means a fixed set of answer options rather than free text.
Three requirements separate a usable question from a decorative one.
Every numeric question needs a unit and a timeframe. “How often do you shop online?” is unanswerable. “In the past 30 days, how many times did you order groceries online?” is not.
Answer options must be mutually exclusive and collectively exhaustive. Overlapping ranges (1–5, 5–10) corrupt the data. Missing options force people into wrong answers. Include “none of these” wherever it could genuinely apply.
The wording must not lead. “How much do you agree that our site is simple to navigate?” has already told the respondent what to think. “How easy or difficult was it to find what you needed?” has not.
Screening and Sample Definition
These come first and determine whether the rest of your data means anything.
1. In the past three months, have you purchased [category]? Yes / No
2. Which best describes your role in choosing [product] for your household? Sole decision-maker / Joint decision-maker / Some input / No input
3. Do you or anyone in your household work in any of the following industries? Market research / Advertising or PR / [your category] / None of these
Question three is a security screener. Industry insiders answer differently from ordinary consumers, and most professional studies exclude them.
Market Sizing and Incidence
4. How many times did you purchase [category] in the last 30 days? Numeric entry, 0–99
5. Approximately how much did your household spend on [category] in the last 30 days? Under $25 / $25–$49 / $50–$99 / $100–$199 / $200 or more / Don’t know
6. Which of the following brands have you purchased in the past 12 months? Multi-select, randomised order, with “None of these”
Randomising brand lists matters more than people expect. Options near the top of a list get selected more often regardless of merit, and that order effect is large enough to change your conclusions.
Behaviour and Frequency
7. In a typical week, on how many days do you use [product]? 0 to 7
8. The last time you bought [category], where did you buy it? Single select, with “Don’t recall”
9. On a typical day, how much time do you spend on social media? Less than 30 minutes / 30–59 minutes / 1–2 hours / 2–4 hours / More than 4 hours
Note the framing of question eight. Asking about the last occasion produces far more reliable data than asking people to summarise their general habits, because recall of a specific event beats self-assessment of a pattern.
Satisfaction and Loyalty
10. Overall, how satisfied were you with [experience]? 1 Very dissatisfied — 5 Very satisfied
11. How likely are you to recommend [brand] to a friend or colleague? 0 Not at all likely — 10 Extremely likely
12. How easy was it to resolve your issue today? 1 Very difficult — 7 Very easy
These are CSAT, NPS, and Customer Effort Score respectively. Each measures something different: recent experience, advocacy, and friction. Running all three at once is common and usually unnecessary — pick the one that matches the decision you will make with it.
Brand Health
13. Which of the following brands have you heard of? Multi-select, randomised, with “None of these”
14. Which of these would you consider the next time you buy [category]? Multi-select from brands recognised in Q13
15. How well does each statement describe [brand]? Matrix: statements down the side, 1 Not at all — 5 Extremely well across the top
Questions 13 and 14 form a funnel: awareness, then consideration, then usage. Tracked over time, the ratios between them tell you whether a marketing problem is one of visibility or persuasion — a distinction that changes the budget entirely.
Keep matrices short. Beyond about six rows, straight-lining rises sharply and the data degrades.
Pricing
16. If [product] were available at $X, how likely would you be to buy it? Definitely would not / Probably would not / Might or might not / Probably would / Definitely would
17. Van Westendorp price sensitivity — four questions asked together:
- At what price would this be so expensive you would not consider it?
- At what price would it be expensive but still worth considering?
- At what price would it be a bargain?
- At what price would it be so cheap you would question the quality?
18. Which would you choose? Option A at $X / Option B at $Y / Neither
Question 16 needs a caution. Stated purchase intent consistently overstates real behaviour, often substantially. Standard practice is to apply a discount factor to the top two boxes rather than treating the raw percentage as a forecast.
Question 18 is the simplest form of choice-based work. Forcing a trade-off between real alternatives produces better data than asking about one option in isolation, because it mirrors how buying actually happens.
Prioritisation and Segmentation
19. Allocate 100 points across these features according to how important each is to you. Constant sum, must total 100
20. Of the five features below, which is most important to you, and which is least important? MaxDiff, repeated across several randomised sets
21. How many people does your organisation employ? 1–9 / 10–49 / 50–249 / 250–999 / 1,000 or more
Questions 19 and 20 both solve the same problem: rating scales produce results where everything is important. Ask people to rate ten features on a five-point scale and eight will come back above four. Forcing a trade-off — allocating a fixed budget, or picking a best and worst — produces genuine discrimination between options.
MaxDiff is more work to field and analyse, but it is the more robust of the two.
Design Decisions Worth Getting Right
Five points or seven? Either is defensible. Research comparing them has found results highly comparable once scores are converted, so consistency across your own tracking matters more than the choice itself. Never change scale length mid-tracker.
Include a neutral midpoint? Yes, where genuine indifference is plausible. Forcing a side on a question people have no view about manufactures opinion that does not exist.
Watch acquiescence bias. Respondents drift toward agreement, especially when tired. Keep surveys short and consider reversing the polarity of a few statements to catch straight-lining.
Question order affects answers. Early questions frame later ones. Ask unaided or general questions before you show brand names or product descriptions, never after.
Use skip logic. Piping and branching keep surveys short and relevant, which measurably improves completion.
Pilot before you launch. Test with a small sample first. Ambiguous wording is obvious in a pilot and invisible in a spreadsheet of 800 responses.
What This Approach Cannot Tell You
Quantitative questionnaires measure how many and how much. They do not explain why.
They are also weak at predicting genuinely new behaviour. People are poor forecasters of their own future actions, particularly around price and novel products, which is why stated intent is discounted rather than taken at face value.
If you need to understand reasoning, motivation, or an unfamiliar problem space, run qualitative work first and use it to write better closed questions. Quantitative research is excellent at measuring things you already know how to ask about.
Final Thoughts
The most common failure in quantitative research is not a badly-worded question. It is a well-worded question that answers something nobody needed to know.
Start from the decision you intend to make. Work backwards to the number that would inform it. Then write the question that produces that number, with a unit, a timeframe, and answer options that do not overlap. Everything else is formatting.
FAQs
What is a quantitative questionnaire?
A survey made up of closed questions producing numerical data you can count, compare across respondents, and analyze statistically.
What are the main types of quantitative questions?
Multiple choice, rating scales, Likert scales, numeric entry, matrix, dichotomous yes/no, ranking, and constant sum allocation.
How many questions should a quantitative survey have?
Fewer than you want. Completion drops sharply with length, so cut anything you cannot name a decision for.
Should I use a 5-point or 7-point scale?
Either works, and results are broadly comparable. What matters most is using the same scale consistently across waves of tracking.
Can quantitative questions tell me why customers behave a certain way?
No. They measure how many and how much. Pair them with qualitative research when you need to understand reasoning.
