How long should a survey be? Fewer questions, one good open box

Yes, but not in a straight line. SurveyMonkey looked at surveys with 1 to 50 questions, deployed between January 2009 and September 2010. Their finding: "the more questions per survey, the higher the respondent drop-off rate from start to finish." They also found that "the relationship between survey length and drop-off rate is not linear" (SurveyMonkey).

The steepest loss comes early. Drop-off climbs fastest with each added question up to 15. After that, each extra question costs less, because the people still there are the ones willing to keep going.

That data is old and comes from one company's customers, so use it as a shape, not a promise. Your audience and how well they know you will move the numbers.

SurveyMonkey's later guidance gives average completion rates by length (SurveyMonkey):

Read this as a gentle slope, then a drop. Going from 10 to 30 questions costs about four points. Going on to 40 costs six more.

The average hides who leaves. Researchers who studied breakoffs found that "the people who break off are most likely to do so near the beginning of the survey" (Plutowski and Zechmeister). So the opening screen matters as much as the length. Say what the survey is, how long it takes and why you're asking, then start with an easy question.

Both, and minutes are easier to plan around. SurveyMonkey reports that people spend "just over a minute" on the first question, around 21 seconds per question for questions 16 to 25, and 19 seconds for 26 to 30. For surveys over 30 questions, time per question is nearly half of what it is in shorter ones (SurveyMonkey).

That last number is a warning. People speed up when a survey drags, so the answers at the end are quicker and thinner.

The same page says abandon rates rose for surveys that took more than 7 to 8 minutes, with completion falling by roughly 5% to 20%. Treat 7 minutes as a ceiling for a survey sent to people who don't owe you anything.

Tolerance changes with the relationship. Pew Research Center pays its panel members, and its own page says it "takes most people about 15 minutes to complete most of our surveys" (Pew). That works because panelists signed up and get a thank-you for each one. A survey link in a newsletter has none of that.

Typing takes longer than clicking, and people know it. SurveyMonkey found that surveys with 10 open-ended questions had a mean completion rate of 78%, against 88% for surveys with one (SurveyMonkey). That is more than 10 points lost.

So the trade is not "short or long". It is how many boxes you ask people to write in. One is cheap. Ten is a different survey.

The one box is also where the useful part lives. A tick box can only return the options you thought of. A written answer can return the thing you didn't. If you have room for one open question, make it count: how to write open-ended survey questions covers wording, and open-ended vs closed-ended questions covers when each fits.

Start from the decisions, not the topics.

1. Write down what you will change with each answer. If nothing would change, drop the question.
2. Cut anything you already know. Plan, signup date and country are often in your own records. Don't ask people to retype them.
3. Merge near-duplicates. Three versions of "how satisfied are you" are one question asked three times.
4. Replace grids with one question. A grid of eight rating rows is eight questions to the person filling it in.
5. Keep one open box, last or second to last. Put it after the easy questions, and make it about a moment ("What nearly stopped you from signing up?") rather than an opinion.

An example

The survey and numbers below are invented to show the method. They are not from a real survey.

A team has a 14-question survey for people who cancelled. Here is how the cut might go.

That takes it from 14 questions to 3. At roughly 20 seconds each, plus a longer first question, it is a minute or two.

There are three common ways to keep a survey short and still learn something.

• A short form you build yourself. In Google Forms or a similar free builder, you write the 3 to 5 questions by hand. It costs nothing, and the work is in deciding what to cut.
• A longer survey with a panel. If you pay people, as Pew does, you can ask for 15 minutes. That is a research project, not a quick check.
• A tool that reads the written answers. The reason people shy away from open boxes is the work of reading them afterwards. If that cost goes away, you can ask one open question of everyone instead of ten tick-box questions.

Enform writes the survey from a description, you share one link, and it reads every reply and counts what people said. Every count opens to the replies behind it, so you can check a label against the words. That fits the "few questions, one open box" shape. It is not a panel, and it won't find you respondents. For how it differs from a general form builder, see the Google Forms comparison. For a ready short survey, try the customer churn survey template. See pricing for what's available.