How to write a questionnaire

Writing with flow

Completing a questionnaire has a psychological flow to it. Questions can trigger memories and thoughts and associations, so the order and flow of questions is important. A good questionnaire will have a logic sequence to help someone think and recall views and thoughts about the area being researched. A poor questionnaire will jump around, or ask things in the wrong order, or ask about items that seem irrelevant. And poor questionnaires lead to lower response rates and fewer completes.

The Wine Glass idea

Connected with flow, is the idea that a questionnaire will have a wine glass 'shape' - starting with broader topics, then narrowing down to the stem of the glass for the details, before broadening out for general demographic detail. Start with open concepts to identify what is important without prompting, then narrow down. Which of these products have you bought this month? Which of brands of fruit juice have you bought this month (unprompted)? Which of these brands have you bought (prompted)? What do you think about Del Monte Orange Juice? See how this moves from general to very specific to avoid biasing the earlier questions by mentioning the brand.

Be specific

Think about the question "Have you bought a printer?". Does this mean 'ever bought' or 'bought this year'? Does it mean any type of printer, or a laser printer, or an ink jet printer? Is it for home or for work? While the questionnaire should move from general to specific, the questions you use also need to be clear along the way - avoiding confusion. A good way to check is to 'pilot' or pre-test the questionnaire before you release it further. Is there anything not understood?

Avoid overlaps and double meanings

Questions often provide lists or bands, but make sure they are separate. 18-25 and 25-35 - where does someone aged 25 tick? Similarly, some questions have more than one meaning, or more than one potential interpretion - Have you travelled by train or plane this year? Is that yes only if both, or yes if just one answer.

Avoid internal jargon and abbreviations

All businesses have an internal jargon, a list of abbreviations and ideas that people working on product design are intimately familiar with. Customers, however, don't think like product managers. They struggle to remember names of things accurately - was it an XR2 or an XRIIB? Abbreviations and initials need to be spelled out in full, and should be avoided where possible - is MVP 'minimal viable product' or 'most valuable player'? Concepts are not necessarily understood such as 'sustainable manufacture' - so check if the phrase is known and understood before you use it.

Write for the person completing the questionnaire

Skipping jargon and internal language is one thing, but the next step is to write questions that reflect how customers talk and think. Do some qualitative research beforehand, or ask someone to give their own words for what they want to buy. Remember than writing that is too high brow or too technical will put people off, and again reduce response rate.

Watch for representative views and edge cases

A good survey has to reach across the whole of a marketplace to be representative. That means writing for people who aren't necessarily your customers, or who, perhaps, don't share the same views as you. Better health may make sense, but there are still people who don't like visiting a doctor. Make sure you capture the width of views, or those who don't like your ideas will drop out and you'll over-estimate potential take up.

Keep it tight

Long questionnaires have poor response rates, and the quality of the answers decreases with length. Stick to the focus of what you are researching, and avoid adding unnecessary 'interesting but unimportant' questions. Keep a focus on what the answers will tell you. If a question won't tell you anything useful, then don't ask it. For mobile phones, quick questions that can be scrolled through can be faster than pages and pages of one-by-one questions.

Avoid forcing and grids

It's common to force people to give answers to questions. And to have horrible long batteries or grids of statements to rate, with no way out. If you force answers people will drop out. They will also give low quality answers just to get through the grid. It can look great in aggregate, but it won't necessarily reflect real opinions or heartfelt attitudes.

Watch for sensitive areas

Remember that a questionnaire gathers opinions, memories and views and relies on recall and memory and includes social biases (yea-saying or willingness to help). Customers don't necessarily have a view of something you might think is important. They also might not be willing to share their view - for instance on what they might pay for something. Or might give a view that reflects social norms, rather than what they really feel in order to avoid confrontation. Sensitive topics need handling with care, and smarter design than just simple questions.

Say please and thank you

Anyone completing a questionnaire is helping you out. Be clear with instructions and questions, but you can also be polite back. Your questionnaire makes an impression on the customers you ask, so may it look and feel like the type of communication you would want to send to a valued client.

Test and test again

Once you have a questionnaire that you think works, try it out before it goes out wider. Better to discover something doesn't work before you ask a few thousand people to respond.