Most surveys don't fail because of bad questions. They fail because of everything around the questions: a cold start, no sense of progress, questions that don't fit what you're asking, and an abrupt ending. Fix those, and the very same questions will collect far more answers — and cleaner data you can actually act on.
Every good survey has three parts, and each part has exactly one job:
1. The welcome page: earn the click
People decide in about three seconds whether your survey is worth their time. Your welcome page has to answer the questions they're silently asking:
What is this about? One sentence on the topic, so no one feels ambushed. "A few questions about your last delivery" is enough.
Why should I care? Say what you'll do with their answers. "Your answers shape what we fix next" works much better than "We value your opinion."
How long will it take? Always show a time estimate. "2 minutes, 5 questions" takes away the fear of an endless questionnaire.
Is it safe to be honest — and what happens with my answers? If answers are anonymous, say so right here. If they're not, say who will read them.
Here's what that looks like on a real survey:
And here's the difference structure makes — the same survey with two different welcome pages. One gets skipped, the other gets answered:
Skip the formal corporate tone. Write the way you would ask a customer face to face. And change the button text — "Share my feedback" invites action, "Continue" invites nothing.
2. The questions: one idea at a time
A question page should feel this calm — one question, a few clear options, and a sense of where you are:
A few simple rules reliably get more people to the end:
Start with an easy question. A simple multiple-choice question warms people up. Save open questions (the ones with a text box), sensitive topics, and anything personal for later.
Every question must earn its place. If you wouldn't change anything based on the answer, delete the question. Keep most surveys under 10 questions and under 5 minutes — five questions with a purpose beat fifteen "nice to know" ones. If you genuinely need more, split it into shorter surveys over time.
Show progress. A progress indicator like "2 / 5" tells people the end is near — so they keep going instead of giving up.
Skip what doesn't apply. Use skip logic — a simple rule like "if they answered No, jump over the next section." If someone says they don't use your mobile app, they shouldn't then see four questions about the app. Each respondent gets a shorter, more relevant survey, and you get cleaner data without "N/A" answers cluttering your results.
Pick the right question type
Using the wrong question format is one of the most common survey mistakes — it makes the survey harder to fill in and the results harder to read. Match the type to what you're actually measuring:
You want to know… | Use this |
How satisfied someone is | Rating scale (1–5 or 1–10) |
Whether they agree with a statement | Likert scale (Strongly agree → Strongly disagree) |
Whether they'd recommend you | NPS (0–10) |
Which option they prefer | Multiple choice |
Why — open feedback | Open text |
Don't default to open text for everything — it's harder for people to fill in and harder for you to analyze. Save it for the follow-up "why?" right after a rating.
Write for someone who's half-distracted
Your respondents are busy — answering between meetings, on a phone, with half their attention elsewhere. If they have to re-read a question to understand it, many will just close the tab. So:
Use plain, everyday language
One idea per question
Short sentences
Avoid double negatives like "How unlikely are you to not recommend…"
Same question, two ways of asking it:
Don't make everything required
Marking every question as mandatory sends a message: "you can't leave until you answer everything perfectly." Mark a question required only when a missing answer genuinely breaks your data. For everything else — especially demographic, optional, or sensitive questions — let people skip, and always offer a "Prefer not to answer" option for anything personal.
3. Make it look worth answering
People judge your survey before they read a single question. A survey that looks clean and on-brand feels trustworthy; a default, unstyled one feels like spam. Your colours, your logo, and a tidy layout aren't decoration — they're the reason someone decides to take you seriously instead of closing the tab.
4. The thank-you page: don't waste the goodwill
The thank-you page is the most underused screen in survey design. Someone just gave you two minutes of their life — use the moment:
Tell them what happens next: "We read every answer. Improvements ship monthly."
Give them somewhere to go: your news page, your help center, a related offer.
If they agreed to a follow-up, tell them when they'll hear from you.
People who see that their feedback leads to real changes are much more likely to answer your next survey too.
Quick checklist before you hit "send"
✅ Welcome page says what it's about, why, how long, and what happens with the answers
✅ Easiest question first, personal questions last
✅ Each question type matches what you're measuring
✅ Wording is plain — one idea per question, no double negatives
✅ Only data-critical questions are required; "Prefer not to answer" where it's personal
✅ Progress indicator is on
✅ Skip logic hides questions that don't apply
✅ Your colours and logo are set
✅ Thank-you page says what happens with the answers