By the Wavefield Research team · Published Sep 29, 2026
A questionnaire is a structured set of questions with fixed wording, response options, and order, used to collect standardized information from respondents. It is the instrument; the survey is the whole project around it — who gets asked, how they are reached, and how answers are analyzed. That distinction, and what makes a good questionnaire, is this page.
The two words blur in everyday use — “fill out this survey” and “fill out this questionnaire” describe the same five minutes — but in research practice they name different things. The questionnaire is the measurement instrument: the questions, their exact wording, their response options, their order. The survey is the study that deploys it: defining the population, drawing the sample, choosing the collection method, fielding, weighting, and analysis. A survey questionnaire is simply the instrument at the center of that machinery.
The distinction earns its keep because the two fail independently. A beautifully written questionnaire fielded to the wrong people — a customer study answered mostly by your happiest customers, say — produces confident, well-measured answers to a question nobody asked. And a perfect sample answering a leading, double-barreled instrument produces garbage at scale. Question and questionnaire quality is one discipline; sampling is another; a defensible study needs both. This page covers the instrument half.
Every item has a job. A good questionnaire is designed backward from the analysis: before writing question one, you know the report's tables, and every item exists to fill one. Questions that no crosstab or decision will ever use — asked because they seemed interesting — cost completion rate and buy nothing.
Standardization is the whole point. Fixed wording and fixed options are what make answers countable. The craft of the individual question — neutral stems, balanced scales, one thing per question — is its own discipline, catalogued with before-and-after rewrites in bad survey questions, and the menu of formats to choose from is in survey question types. Pew Research Center's Writing Survey Questions is the reference on how much wording and question order move answers — the same opinion, asked two ways, routinely yields different numbers.
Order is a design decision. Easy and engaging items first, the headline measure before the detail that could contaminate it, sensitive items and demographics last, and screeners up front. Length is too: completion quality decays measurably past ten minutes, and every marginal question taxes every answer after it.
A complete café customer questionnaire — deliberately ordinary subject matter, because the structure is the lesson. These questionnaire items are examples of a pattern that transfers to almost any feedback study: classify, anchor to behavior, measure the headline, diagnose, detect problems, close with demographics and a safety valve.
1. How often do you visit us?
Single choice: First visit · Less than monthly · 1–3 times a month · Weekly or more
The classification item. Every later answer gets read against visit frequency, so it comes first and its bands match how results will be reported.
2. What did you order today?
Multi-select from the menu categories + Other
Behavioral, not attitudinal — what people did, not what they think. Anchors the feedback to a concrete transaction.
3. How satisfied were you with your visit today?
5-point scale, Very dissatisfied to Very satisfied
The headline measure. One scale, labeled at every point, asked before the detail items so their wording cannot contaminate it.
4. How satisfied were you with each of the following?
Short matrix: speed of service · staff friendliness · drink quality · value for money
The diagnostic block. Four rows maximum — every added row costs completion quality — chosen because each maps to something the owner can act on.
5. Was anything about today's visit frustrating?
Yes / No, with a follow-up text box on Yes
The problem detector. Gating the open end behind a Yes keeps the text box for people with something to say.
6. How likely are you to recommend us to a friend?
0–10 scale
The loyalty measure, with its own follow-up asking the main reason for the score — the verbatims carry more signal than the number.
7. Which age group are you in?
Census-aligned bands, plus Prefer not to say
Demographics live at the end, after goodwill is earned. Bands are chosen to match the crosstab plan, not invented.
8. Anything else you'd like us to know?
Optional open end
The safety valve. Always optional, always last — and reliably the source of the one theme nobody thought to ask about.
For a professional-grade benchmark, read a government instrument: the U.S. Census Bureau publishes its American Community Survey forms in full — decades of wording refinement visible on every page, and the source of the census-aligned answer options that commercial demographic questions should copy.
Questionnaires are typed by how they reach the respondent, and the type sets the design rules. Online — self-completed on a device — is the default: cheapest to field, fastest, and the only type where logic, piping, and randomization come free. Paper and postal survives where devices don't reach or a physical moment matters (the comment card, the clinic intake form). Telephone- administered puts an interviewer's voice on every question, which rescues complex studies and hard-to-reach audiences at real cost per complete. Face-to-face adds show cards, probing, and observation — the gold standard for some government and qualitative work, and the most expensive by far.
The rule that falls out: the less help a respondent has, the simpler the instrument must be. A self-completed questionnaire cannot clarify, so it needs shorter questions, plainer words, and stricter structure than anything an interviewer reads aloud. Note this taxonomy is about the instrument's delivery — the question-level taxonomy (multiple choice, Likert, ranking, open-end) is a different axis, covered in survey question types.
Starting a questionnaire does not begin with writing questions. The working sequence: name the decision the study serves and the two or three things it must find out; sketch the report — the actual tables and charts you intend to show, because each one implies the items that feed it; draft the instrument — screeners first, easy opener, headline measures early, diagnostics in the middle, demographics last, ten minutes or less; then pilot it. A pilot of even ten people catches the question everyone misreads, the option list missing its most common answer, and the honest completion time — faults that are cheap before fielding and unfixable after. The craft of qualitative slots inside a mostly-quantitative instrument is covered in qualitative survey questions.
On Wavefield, this sequence is what the AI agent runs when you hand it a brief: it drafts the instrument from your research goal — structure, wording, census-aligned options, logic — and you review and edit before anything fields. The soft-launch step then works as a built-in pilot: field the first twenty completes, check the data, and only then open the study fully.
A questionnaire is the structured set of questions used to collect information from respondents — the instrument itself, with fixed wording, response options, and question order, so every respondent answers the same thing the same way. The definition of a questionnaire is about standardization: it is a measurement tool, the way a thermometer is, and its quality determines whether the answers can be compared and counted.
The questionnaire is the instrument; the survey is the whole project that uses it. A survey includes deciding who to ask (sampling), how to reach them, fielding, and analysis — the questionnaire is the question set at the center of it. In everyday speech the words blur, and 'survey questionnaire' is common; in research practice the distinction matters because a good questionnaire inside a badly sampled survey still produces wrong answers.
Questionnaire, with two n's. The word comes from French, where -aire endings follow a double n in this case. 'Questionaire' is a common misspelling that spell-checkers usually catch.
By how they are administered: online (self-completed on a device, now the default), paper or postal (self-completed by hand), telephone-administered (an interviewer reads the questions), and face-to-face (an interviewer present, allowing show cards and probing). The type changes design rules — an interviewer can clarify, a self-completed form cannot, so self-completed questionnaires need simpler wording and stricter structure.
The annotated café customer questionnaire on this page is a complete example: eight items running from a classification question through satisfaction measures, a diagnostic matrix, a problem detector, a loyalty question, and closing demographics. For a professional benchmark, the U.S. Census Bureau publishes its American Community Survey forms — a masterclass in precise wording at national scale.
Research that collects standardized data by having respondents answer a fixed instrument, as opposed to interviews or observation. Its strength is comparability and scale: hundreds or thousands of people answering identical questions produce countable, crosstab-able data. Its limit is that it can only measure what it thought to ask — which is why exploratory work often runs qualitative methods first and the questionnaire survey method second.
Related: survey question types · bad survey questions, fixed · qualitative survey questions · data collection methods
Describe the decision you need to make and the agent drafts the instrument — structure, wording, options, logic — for you to review, edit, and field. From $99 per study.
Start a study