How long should a survey be?

As short as the decision behind it allows. That sounds like a dodge, but it is more useful than any number: length is a cost the respondent pays, so the right length depends on what you need to learn and on what the people answering are willing to give.

This article turns that into a method. Measure length in minutes, set a budget that fits your audience, cut to it, and check the real length once answers arrive.

Count minutes, not questions

“Ten questions” can take one minute or fifteen. A multiple-choice question is read and answered in seconds. A rating grid with twelve rows is twelve answers, not one. An open question asks someone to compose a reply, which takes far longer than picking an option.

So estimate in time. Add up the reading (instructions, descriptions, any passage people must read before answering) and the answering, counting each grid by its rows and giving each open question real time. The free completion time estimator does this arithmetic from your counts of each question type and the number of words people will read.

Two things keep an estimate honest. First, time someone else, not yourself: you wrote the questions, so you read them faster than anyone else will. Second, estimate the route people actually take. With branching, a respondent may see only part of the survey, so the length that matters is the longest likely route, not the total number of questions.

Allow for the device, too. On a phone, typing takes longer and a wide grid turns into a long scroll, so open questions and matrices cost more there than they do at a desk. If most of your audience will answer on a phone, time the survey on one.

Set the budget by who is answering

There is no universal number, because the same five minutes are cheap in one context and expensive in another. Four questions set the budget:

• What do they get? People paid to take part in a study have agreed to give their time. A customer who received an email has agreed to nothing yet.
• Where are they? Someone answering on a phone between other things will give you less than someone at a desk who set time aside.
• How much do they care? People give more time to surveys about things that affect them: their own team, a service they depend on, a decision that will change their day.
• How often do you ask? A short survey every month can use up more goodwill over a year than a longer one sent once.

Write the budget down in minutes before you write any questions. It turns every later request to add “just one more question” into a trade: what comes out to make room?

Rough shapes for common surveys

Budgets differ, but some shapes come up again and again:

• A question on a website or in a product. One question, perhaps two, answered without leaving what the visitor was doing. Anything longer is a different survey and deserves its own invitation.
• After a support conversation or a purchase. A score, the reason for it, and perhaps one question about what would have helped. The experience was small, so the survey should be too.
• A periodic customer or member survey. Longer is reasonable, because it covers the whole relationship, but every section should connect to a decision coming up in the months ahead.
• An employee survey. People answer at work, about something that affects them, which buys more time than a customer survey gets. That goodwill lasts only as long as earlier surveys visibly led somewhere, and short regular check-ins are a different design again; see employee pulse surveys.
• A paid research study. As long as the design needs, stated honestly and paid fairly, with checks that the extra minutes are not being spent speeding.

These are shapes, not rules. The four questions above decide the budget for your survey.

Cut with the decision test

For each question, ask what you will do differently depending on the answer. If the honest answer is “nothing”, the question is interesting rather than useful, and interesting questions are the first to go.

Other cuts that rarely cost anything:

• Questions you can answer from data you already hold, such as plan, sign-up date or department.
• Two questions that measure the same thing in different words.
• Demographics you will never use to split the results. Ask for the groups you will actually compare, and no more.
• An open question asking for something a short list of options would capture.

If a question survives but only matters to some respondents, route everyone else past it instead of asking all of them. The article on planning skip logic shows how to map that before you build it.

What length costs you in the data

Long surveys do not only lose people. The people who stay start answering differently: faster, with less thought, with the same answer all the way down a grid. A matrix question late in a long survey is where straight-lining tends to appear, and a respondent who speeds through to the end still counts as a completed response.

That makes length a data quality problem as much as a completion rate problem. Answers to your last questions come from a tired, self-selected subset of the people who started. The guide to catching low-quality responses covers how to spot the patterns that result.

Make the time easier to give

Some surveys are long for good reasons. You can still make the time easier to give:

• Say how long it takes, honestly, before the first question. People who know what they are agreeing to are less likely to feel ambushed halfway through.
• Put the questions that matter most early, while attention is fresh, and keep demographics for the end.
• Group questions into short pages with clear titles, so progress feels steady rather than endless.
• Start with an easy, relevant question, not a wall of instructions or a question about income.

The format matters too. For a short survey, showing one question at a time can make the whole thing feel lighter; for a long one, a page of related questions can be quicker to work through. The comparison of conversational and classic surveys covers when each helps.

When a long survey is the right call

Detailed research instruments, validated scales and studies with paid participants can justify a longer questionnaire, as long as the time is stated up front and, where people are paid, the payment reflects it.

If you need more than your audience will give in one sitting, split the work. One option is a short core that everyone answers, with extra modules that different people answer. Another is separate waves spread over time, each short enough to finish comfortably. In Plumeform, surveys can be grouped as waves of a longitudinal study on Pro, with answers linked by a participant ID; multi-wave studies explains how that works and how to plan for people who drop out between waves.

Check the real length after launch

Your estimate is a plan; the first responses tell you what the survey actually costs. If your survey has a welcome screen, Plumeform shows respondents an estimated time on it, worked out from the question types and the amount of reading, unless your own welcome text already says how long it takes. Once answers arrive, the results summary shows the median time to finish and lists the questions where people left.

Read both early. If the median time is well over your budget, cut before collecting more. If many people leave at one question, read that question again: it is usually too long, too personal, or unclear about what it wants. The pre-launch checklist includes sending to a small first batch so you can make those fixes before most of your audience sees the survey.