A questionnaire maker, and how to not waste the responses

Strip away the templates and themes and four features decide whether the data is usable. First, branching (also called skip logic or conditional logic): the ability to show a question only when an earlier answer makes it relevant. Without it, people answer questions that do not apply to them and pick the least-wrong option, which quietly contaminates your results.

Second, a response cap you will not hit mid-survey. A free plan that stops accepting answers at 10 or 100 responses is fine for a class project and a problem for a customer survey. Third, partial-response or drop-off data, so you can see the question where people quit. Fourth, a clean export (CSV or a live Google Sheets connection) so analysis happens in a tool built for it.

Everything else — themes, AI question generation, logo placement — is nice to have. Check the four above first, then compare the rest.

If you need unlimited responses at zero cost and your questionnaire is simple, Google Forms is hard to beat, and Tally adds a more flexible editor with no stated cap. Pick Google Forms with care if respondents must upload files, because each of them has to sign in to a Google account first.

If you need 10 questions or fewer, no branching and a handful of responses, SurveyMonkey's and Zoho's free tiers work. For anything larger, look at the response cap before the feature list. The sections below apply whichever tool you choose.

Start with the sentence you hope to write at the end: "X percent of Y said Z, so we will do W." Write those sentences before you write a single question. Every question should feed one of them; anything that does not is costing you respondents without paying you back.

Next, write each question to measure one thing in words every respondent will read the same way. Pew Research Center's guidance shows how much wording matters: in one Pew survey, 68% favored military action in Iraq, but when the question added that U.S. forces might suffer thousands of casualties, support fell to 43%. Same topic, different words, a 25-point swing.

Then decide the order deliberately. Earlier questions frame later ones, so put general questions before specific ones on the same topic, keep sensitive or demographic questions near the end, and put any screener at the very start. Finally, pretest. AAPOR's best-practice guidance recommends pretesting with people similar to your real respondents, typically through cognitive interviews where they explain how they understood each question.

Shorter than you think, and measured in minutes rather than questions. SurveyMonkey's analysis of about 100,000 surveys found respondents spend roughly 5 minutes on a 10-question survey and that abandonment rises for surveys taking more than 7–8 minutes. Our guide on how many questions a survey should have traces that and other commonly quoted figures to their sources.

Before you send, run your question mix through our survey length calculator at /tools/survey-completion-time-calculator to get an estimate you can put in the invitation.

Zunoform's free plan gives you unlimited questionnaires, 500 responses a month, conditional logic, answer piping, hidden fields, partial-response saving and an anonymous mode that stores no IP address or device data. You can run the same questions in a conversational, single-page or document layout, and translate a questionnaire into several languages.

It is not the right pick if you need unlimited free responses (Google Forms and Tally offer that), a decade-deep integration catalog (Typeform and Jotform are ahead), or academic survey features such as panel sourcing, quotas and built-in statistical analysis — dedicated research platforms like Qualtrics are built for those. AI questionnaire generation exists but is on the Business plan only, and publishing NPS, signature or payment questions needs Pro.