How many questions should a survey have, and what actually sets the limit?
There is no correct number of questions, and the numbers circulating online come from a different kind of survey entirely. What sets the ceiling is how much effort your questions ask for, not how many of them there are.
In brief
Most quantitative consumer studies run 20 to 30 questions inside a 10 to 12 minute budget, but the count is derived rather than chosen. Set a target completion time, price each question by the effort its type demands, and fit as many as the budget allows. A 12-question study built on long grids and open text asks more of a respondent than a 30-question study of single-choice questions.
Why question count is the wrong measure of survey length
The guidance in circulation puts the answer somewhere between five and twenty questions, and it disagrees with itself because it is answering a different question. Those figures come from self-serve web forms sent to unincentivised audiences: feedback requests, event forms, post-purchase surveys where the respondent has no reason to finish. In that context a twelfth question really does cost you completes.
Panel-based consumer research works differently. Respondents opted in, they are incentivised, and they expect a study rather than a form. Studies of 25 to 35 questions complete routinely. Importing the five-to-ten rule into commissioned research would make most category work impossible.
The unit that matters in both cases is effort, not count. Two questions can differ in cost by a factor of six. A 5-point rating takes a few seconds. A ten-row grid takes over a minute and demands sustained attention across every row. Counting them as one each tells you almost nothing about what you are asking of a respondent.
How to set a survey length budget in minutes
Set the completion time first, then fill it.
Working budgets for consumer research run 2 to 3 minutes for a pulse check, 6 to 8 for a single-focus test, 10 to 12 for a standard study, and 15 or more only where the audience has a strong reason to engage and the incentive reflects it. Past 15 minutes you are trading data quality for question count, whatever the completion rate says.
Working backwards from minutes forces the right conversation. “We have twelve minutes, these questions cost fourteen, what comes out” is a solvable problem. “How many questions can we have” is not, because it has no answer anyone can defend and the list grows until someone stops it arbitrarily.
How long each survey question type takes to answer
Approximate respondent time per question, useful for planning rather than precise measurement. Times run longer on mobile and longer again for audiences answering in a second language.
| Question type | Approximate time | Notes |
|---|---|---|
| Single choice, short list | 10 seconds | Cheapest question available |
| Rating or slider | 8 to 10 seconds | Fast once the scale is understood |
| Likert scale, single row | 8 to 10 seconds | Same as a rating |
| Multiple choice | 15 to 20 seconds | Scales with option count |
| Image choice | 20 to 30 seconds | Respondent has to actually look |
| Likert grid | 10 seconds per row | A ten-row grid is a minute and a half |
| Ranking | 30 to 45 seconds | Drag interaction is slow, especially on mobile |
| Short answer | 20 seconds | Typing is the bottleneck |
| Open text | 45 to 60 seconds | The most expensive question in any survey |
| Typed list | 40 seconds and up | Depends on how many items you require |
Grids and open text usually account for most of a questionnaire's runtime while making up a small share of its question count, which is why counting questions misleads. It also means a study can be shortened substantially without cutting anything from it. Converting a twelve-row grid to six rows, or an open question to a closed list, buys back more time than deleting three single-choice questions.
How piping changes the question count
Piping repeats a question once per item the respondent selected earlier. That means your question count is not a number, it is a range.
A follow-up piped from a multiple-choice question where respondents select between one and six items is one question in the builder and up to six in the interview. A respondent who selected six sees a materially longer survey than one who selected one, and if the piped question is an open text you have just handed the first respondent six minutes of typing.
Design for the worst case. Estimate the average selections from the source question, plan the budget on that, then cap the loop, commonly at three, so the heaviest respondents are not answering a different survey from everyone else. Where the cap bites, pipe the items that carry the decision rather than everything the respondent chose.
Survey length by study type
| Study type | Questions | Target minutes |
|---|---|---|
| Pulse or single-issue check | 4 to 8 | 2 to 3 |
| Price sensitivity | 8 to 15 | 5 to 8 |
| Packaging or name test | 10 to 20 | 6 to 8 |
| Concept test | 15 to 25 | 8 to 10 |
| Brand health or tracker | 20 to 30 | 10 to 12 |
| Category deep dive | 25 to 35 | 12 to 15 |
Trackers sit at the shorter end of their range for a reason. They run repeatedly against the same audience, so length compounds across waves, and a tracker that is tolerable once becomes a drop-off problem by wave four.
What happens when a survey is too long
Overlong surveys rarely fail visibly. They return complete responses that are worse in the second half than the first.
Analysis of around 100,000 surveys by SurveyMonkey found abandon rates rising for surveys that ran beyond seven to eight minutes, and respondents to surveys over 30 questions spending close to half as long per question as those answering shorter ones. That second finding is the important one. The responses still arrive. They are just less considered, and nothing in the dataset marks which ones.
The damage is specific and none of it is flagged. Open-text answers shorten towards the end of a long survey, which is why the final open question returns single words while the first returns paragraphs. Grid responses flatten into straight lines, and a flat grid reads as a considered attitude rather than as fatigue. Drop-off compounds both, because the respondents who abandon a long survey are not a random subset of the audience.
How to shorten a survey questionnaire
Start with decision relevance. Any question whose answer would not change an action comes out, regardless of who asked for it.
What survives can usually be converted rather than deleted. A twelve-row grid becomes six rows, keeping the attributes that differentiate. An open question becomes a closed list built from answers you have already seen. A broad multiple-choice question becomes a single choice where only the primary answer matters.
Logic does the rest. Questions that apply to a subgroup should be shown to that subgroup rather than to everyone with a “not applicable” option, and conditional reveal shortens the path each respondent walks without narrowing what the study covers.
Whatever remains gets reordered so the load-bearing questions sit in the first half. If the study has to run long, the length should fall on the measures you can afford to lose precision on.
Splitting into two studies is the last resort and occasionally the right one. It costs a second sample, so it only makes sense when both halves need full cohort bases anyway.
Survey length rule of thumb
If you want one figure, hold 10 minutes rather than a question count. Estimate the questionnaire against the effort table, and if it exceeds 10 minutes, something comes out or converts. That single constraint resolves most arguments about length, and unlike a question count it holds regardless of what types the questionnaire uses.
Length checklist
Minutes budget set before the question list. Each question priced by type, not counted. Grids and open text audited first, since they carry most of the runtime. Piped questions capped and estimated on average selections. Load-bearing questions in the first half. Logic used to shorten each respondent's path. Estimate validated against pilot timing before fielding.
Why Flickly?
Flickly builds quantitative studies with question types, piping and conditional logic configured against a target length, then measures actual completion time against your estimate on every response.
Frequently asked questions
Run your next study with verified Indian consumers
Design the study, reach the right respondents, and get decision-ready insights in 72 hours.