There’s no universal answer. Anyone who gives you one (“keep it under 10 questions!”) is selling you simple advice that won’t survive contact with your specific survey.
The right length depends on four things: who your audience is, what they’re getting in return for answering, how the survey reaches them, and what kinds of questions you’re asking.
This article gives you the decision rule, real data on what actually happens to completion rates as surveys get longer, and the most useful skill of all — how to cut questions you think you need.
The honest answer: it depends on 4 variables
1. Audience motivation. A paid panelist who signed up to take surveys will tolerate 30 minutes. An anonymous customer answering an email link will tolerate 3 minutes. An employee responding to an HR survey on company time sits somewhere in between.
2. Incentive. No incentive: keep it under 5 minutes. A $5 gift card: 10 minutes is fair. A $50 incentive on a panel: 25 minutes is fine. Incentive elasticity is real and roughly linear up to a point.
3. Channel. Email survey link: 3–5 minutes is the realistic ceiling. SMS link: 90 seconds. Embedded in-product survey: 60 seconds. In-person tablet/iPad: 5–8 minutes. Each channel has a different attention budget.
4. Question type. Multiple-choice questions take 5–8 seconds each. Likert items take 6–10 seconds. Open-ended questions take 30–60 seconds each, and they crater completion rates. A “10-question survey” with three open-ended questions takes longer than a “20-question survey” with none.
A 15-question survey targeting unpaid customers via email is short. The same survey targeting paid panelists with a $20 incentive is offensively short. Same survey, different math.
What actually happens to completion rates as surveys get longer
We pulled this from our own data rather than citing someone else’s: 400,050 real survey responses on SurveyRock over the last three years, filtered for data quality (no test accounts, no bot-speed completions, no broken timestamps — full methodology in the endnote). Here’s what completion rate looks like by question count:
| Questions | Completion rate |
|---|---|
| 1–5 | 99.4% |
| 6–10 | 97.4% |
| 11–15 | 98.0% |
| 16–25 | 77.1% |
| 26+ | 80.3% |
The headline: completion rate falls off a cliff between 15 and 16 questions — from 98% down to 77%, a 21-point drop in one bucket. Below 15 questions, completion barely moves at all; the curve is nearly flat.
Median time-to-complete tracks the same pattern: 28 seconds for a 1–5 question survey, 3.25 minutes for 11–15 questions, and 9.6 minutes once you cross into 26+ questions.
Two findings worth knowing before you plan a long survey:
Most abandonment happens before the halfway point, not near the end. Looking at where people who don’t finish actually stop: two-thirds of all abandonment (67%) happens in the first half of the survey. If a respondent makes it past 50%, they’re considerably more likely to finish than to quit. This changes where you should spend editing effort — the first half of a long survey is where you’re losing people, not the last few questions.
Mobile respondents fall further behind on long surveys specifically. On surveys under 15 questions, mobile and desktop complete at roughly the same rate. Past 15 questions, mobile completion drops 6–8 points below desktop (75.7% vs. 82.0% in the 16–25 bucket; 76.2% vs. 84.0% in the 26+ bucket). If your audience skews mobile — and for most distribution channels in 2026, it does — the 15-question line matters more for you than for a desktop-only audience.
Open-ended questions cost real completion. Surveys with at least one open-ended question complete at 90.8%, versus 95.1% for closed-ended-only surveys — a 4.3-point gap, consistent across the dataset. Not a reason to avoid open-ended questions (they’re often the most valuable data in the survey), but a real cost to budget for.
The decision rule
A rough table for picking a length range:
| Audience | Incentive | Channel | Target length |
|---|---|---|---|
| Active customers | None | 3–5 minutes (10–15 questions) | |
| Active customers | $5–10 | 5–8 minutes (15–25 questions) | |
| Employees (work time) | None | Email/Slack | 8–12 minutes (20–35 questions) |
| Churned customers | $10–25 | 8–15 minutes | |
| Anonymous web visitors | None | Pop-up / link | 60–90 seconds (3–5 questions) |
| Paid research panel | $20+ | Panel platform | 15–25 minutes |
| In-product feedback | None | In-app | 30–60 seconds (1–3 questions) |
| SMS survey | None | SMS | 30–60 seconds (1–3 questions) |
These are starting points, not commandments. Use your own drop-off data to adjust.
How to cut questions you think you need
This is the actual skill. Cutting questions from a survey draft is harder than writing them, because every question feels important when you wrote it.
The test that works:
For every question, ask: what decision will I make differently based on the answer?
If you can name a specific decision — keep the question. If you can’t — cut it.
Most “nice to know” questions fail this test. They get added because someone in a meeting said “could we also ask about…”. Could you? Yes. Should you? Only if the answer changes a decision.
A few patterns worth watching for:
- Demographic questions you don’t segment by. If you ask company size but never analyze results by company size, drop the question.
- Questions that are also in your CRM/HRIS/database. Don’t ask respondents what you already know.
- Multiple questions measuring the same thing. “How satisfied are you?” and “How would you rate your experience?” and “How happy are you?” are all the same question. Pick one.
- “Anything else?” open-ended questions early in the survey. Move them to the end. Putting them earlier just gives respondents an off-ramp.
Measure, don’t guess
The single best thing you can do is stop guessing about length and start measuring drop-off on your own surveys.
If you’re seeing high abandonment at question 7, the problem isn’t “the survey is too long.” The problem is question 7 — either it’s confusing, it asks for information the respondent doesn’t have, or it feels invasive. Fix that specific question and your completion rate climbs.
SurveyRock tracks a page-level completion funnel — completed, abandoned, and in-progress counts at every step — plus per-question timing (median and 90th-percentile time-on-question), so you can see exactly which question is where people stall, not just that the survey overall has a drop-off problem.
When longer surveys are actually fine
A few cases where length matters less:
- Paid research panels. Respondents signed up for this.
- Academic research. The respondent is participating in something they understand requires depth.
- Highly engaged customers. Your power users will answer a 30-question survey if it shows you’re taking them seriously.
- Once-a-year benchmarks. Annual engagement surveys, deep customer satisfaction studies — when the answers genuinely require depth, the length is justified.
The mistake is treating these cases as the rule rather than the exception.
Bottom line
Length isn’t a rule, it’s a constraint that flexes with audience, incentive, channel, and question type. Use the decision-rule table as a starting point, cut every question that doesn’t drive a decision, and measure drop-off on your own surveys so you can shorten where it actually matters.
The shortest survey is rarely the best one. The most ruthlessly edited one usually is.
Want to see where respondents drop off your survey? Start a free SurveyRock account — the completion funnel and per-question timing are built in.
Methodology: 400,050 real SurveyRock survey responses from the last three years (2023–2026), filtered to exclude test/demo accounts, broken timestamps, and bot-speed completions (see the full methodology and additional findings in our companion piece on completion-rate benchmarks).