How to design a survey—11 tips from the experts

Know what you're trying to find out

Keep this question in mind while you’re making the survey—and then come back to it repeatedly.

Like with any project, it’s important to have clear, smart goals here. If you don’t know the purpose of your survey, then you can’t even think about starting it. When you can answer this question with 100% certainty, then you’re ready to start designing.

If you’re going to ask a question, and it’s not going to help with your core goal, then change it up or take it out.

Keep survey-takers in the loop

Let the people taking your survey know the purpose in the survey introduction. If you don’t feel like reading the whole thing, which you should do later, here are the basics of that topic. The introduction should clarify:

1. Who you are 2. What you want to do with this survey 3. How long this will take 4. If their answers are anonymous 5. And any other relevant instructions

Think about your demographic

You can slice society into tiny fractions with demographic survey questions. There’s gender, profession, age, location, even favorite sports teams. These are often useful—but don’t ignore behavior. Would you ask a vegan how they marinate their sirloin steak? Nope, so keep important details like that in mind when you design your survey.

Remember to test

Testing your survey is vital. But don’t just ask your colleagues—ask regular people, you know, like your friends and family. This can give help you polish rough diamonds and make it easier for your audience to understand. Now if your friends, family, or neighbor’s mother don’t fully understand the questions, or get confused by a lack of context, then the same is likely to happen to your real participants, and your data is going to suffer.

Use qualifying questions

One of the best ways to start a survey. You want to make sure everyone you’re asking can give you useful information. Remember the demographic you’re interested in learning from and make sure those people are the ones giving you answers.

Let’s say you run an online art supply store, and you want to know how often people change their brushes. A necessary screening question would be, “How often do you paint?” If the answer is never, then this isn’t a person whose experiences can help you here.

Try open and closed questions

Think about how you plan to use this data. If you’re planning to make graphs, charts, or infographics that people can quickly see and absorb, then ask closed questions to give yourself clear numerical information.

If you want to learn more about people’s opinions, experiences, and beliefs, then offer open-ended survey questions. Remember, these will take respondents longer to answer than closed questions, to factor that in when designing the survey.

Experiment with other question types

The answers to surveys aren’t always just yes or no. Maybe you want to ask how comfy the hotel beds were on a scale. Or you want to find out what fruit your customers buy every week. Then you can use Likert Scales or Multiple Choice questions. However you decide to approach a question, make sure all the answers are available.

Keep the language human

Ditch the jargon. Don’t force your readers to spend precious brain power doing your job for you—like making things simple and understandable. You want them to absorb the topic easily, and to use their time to think about their answer.

The more you speak like a human, instead of an academic textbook, the better your chances are to take people from Start to Submit.

Ask specific questions

“What’s the best vegetable?”.

What’re your criteria for “best”? The tastiest? The easiest to cook with? The most colorful? The most nutritious?

When it comes to surveys, specificity is king. Otherwise, people won’t know what to answer, get bored, and your data will suffer. If you’re interested in all of those four elements of what makes a great vegetable, then ask them in four separate questions.

Likewise, you want to avoid asking multiple questions. If you ask, “what is the most interesting and exciting film ever?”, then your readers are going to want to put two answers down. So either ask those two questions separately or pick one and stick with it.

Don't ask loaded questions

You’re not squeezing answers out of a mobster in an interrogation room. You’re speaking to people who have volunteered their time to answer questions. So don’t trick people into giving you the answers you want.

Avoid using assumptive, leading questions. If you ask, “How many goals will Springfield FC win by this weekend?”, there will probably be people who don’t think they’ll win at all. But here you’ve left them with no option but to agree that they’ll win. Instead, ask “What will the score of the game be?”

Think about the order of your questions

Once you have your list of questions selected, it’s time to start asking.

Or not. You can’t just throw a randomized list of questions at people. The structure is important, too.

Start with your screening questions, because there’s no point making someone sit through a questionnaire, only to be rejected at the end. It’s just nice to do.

All in all, there should be a logical flow to questions. Questions of the same topic should be kept together. If you change topics a few questions into the survey, take them back to the surface again. Start broad, end specific.

Finally, it’s better to leave open-ended questions to the end. By this point, your respondent will have been thinking about the topic at hand for a few minutes, so you can expect better-quality answers.