Long-survey UX: where people drop out
Every extra question costs you 3–7% of completions. How graph logic, an honest progress bar and better wording keep respondents to the end.
Stakeholders love long surveys. “While we’re at it, let’s also ask about age, city, income, favourite colour and years of experience.” Every one of those breaks the funnel. Here are the numbers on where respondents actually drop out, and what to do about it.
How many questions a respondent will tolerate
From our data (aggregated across Askyo launches in 2025–2026):
| Survey length | Median completion rate |
|---|---|
| 1–3 questions | 87% |
| 4–7 questions | 71% |
| 8–14 questions | 52% |
| 15–25 questions | 31% |
| 25+ questions | 12% |
These are medians. If the survey is anonymous, incentivised and optional, you do better. If it is mandatory after signup, you formally get 100%, and random checkboxes instead of data.
The takeaway is simple: every three extra questions cost you 5–10% of completions.
Where exactly they leave
A heat map of drop-off by position:
- Question #1 – 5–8% leave immediately, before answering. Accidental clicks, plus people who saw the word “survey” and closed the tab.
- Questions #2–4 – a flat plateau, 1–2% per question.
- Question #5 – a small spike, about +3%. This is “already the fifth one, how many more?”
- Questions #8–12 – the fatigue zone. 2–4% per question.
- The personal-data question – a peak of its own, +5–8% right there, especially with no explanation of why you are asking.
- The last 2–3 questions – a small recovery bump. People who can see “almost done” finish.
Eight things that actually work
1. Graph logic instead of a linear list
Having 25 questions does not mean every respondent must answer all 25. With a graph editor you branch: “not employed → skip the eight questions about work”. Each person sees exactly what is relevant to them.
In our data, surveys with conditional logic retain 18–24% more people at the same nominal length.
2. A progress bar, but an honest one
“Step 3 of 12” works. “Step 3 of 47” kills. If the survey is long, show a percentage rather than an absolute count. And do not lie: “95% done” with ten questions still ahead is something people feel, and they close the tab.
3. One question, one screen
A wall of 30 fields is intimidating. Show one question at a time with plenty of space: it feels shorter psychologically even when the total is identical.
In Askyo that is the one-by-one runner mode instead of the scrolling form.
4. Open questions go at the end
Typing takes effort. An open question early on hits people before they are invested, and many will leave. At the end, they are too far in to abandon.
5. Explain why you are asking
“Enter your email” reads as spam bait. “Enter your email so we can send you the study results, one email, not a subscription” converts two to three times better.
6. Let people skip what is not essential
If a question genuinely is not critical, make it optional. Better to get eight answers out of ten than to lose the respondent at question nine.
7. Keep the wording neutral
“What problems do you have with our product?” presupposes there are problems. Better: “What would you change about our product if you could?” It is open, presumes nothing and gives the respondent room.
8. Test on 30 real people
Before you launch to 5,000, take it yourself, give it to three colleagues and show it to five friends outside the team. Every awkward phrasing, unexplained term and broken branch will surface.
A concrete example: a SaaS onboarding survey
Before: 18 questions, linear. Completion rate 38%.
After the rewrite:
- 6 baseline questions for everyone.
- 4 branches of 3 questions each depending on role (developer / marketer / founder).
- Email and contact optional, at the end, with an explanation that we will send personalised recommendations.
- A progress bar reading “1 of ~10” instead of “1 of 18”.
Completion rate went to 67%. The average number of questions per respondent dropped to 8.4. Data quality did not suffer, it improved, because there were far fewer fatigue-driven random answers.
A checklist for surveys that do not exhaust people
- You wrote an opening question that earns attention before building anything
- Graph logic is on, so irrelevant branches stay hidden
- The progress bar is honest: a percentage or a realistic count
- Open questions sit at the end
- Every optional field explains why you want it
- The survey has been taken by hand at least five times, by different people
- Total length stays under 12 minutes (roughly 12–18 questions)
The Askyo editor estimates completion time automatically as you build. Nine minutes is fine. Twenty-three minutes means it is time to cut.