Tactical survey design can mean the difference between a great survey experience and a truly terrible one.
So, you’ve decided to run a survey. You are confident it’s the best way to gain insight into your research project. You sit down to write the questionnaire, but you’re stuck. You know the questions you want to ask, but don’t know how to ask them. You’ve heard that survey design is important, but don’t know where to start.
Luckily, there are a few basic principles that can elevate your survey from solidly mediocre to just plain solid.
Here are eight top tactical design tenets to remember when running a survey:
- Clearly Define Your Objective: Your objective is your guiding star. It informs how you structure your questions, the specific language you use, and ultimately what data you get at the end. With your objective in mind, you can focus on what exactly the survey needs to accomplish. Typically, the goal of a survey is to collect data that can be analyzed for actionable use. If your data doesn’t enable confident action, then the survey has failed its objective.
- Always Include a Screening Section: Whether you are surveying CEOs or consumers, a screening section will ensure that you are getting the right people to answer the right questions. What are the defining factors that make an ideal respondent? What is their role, their seniority, their experience with a given product? As a rule, screening questions with Yes/No answer choices should be avoided. Instead, offer a range of potential answers and allow the respondent to qualify by selecting the correct response. For example, instead of asking “Are you a CEO?” you could ask “What is your role?” and have CEO listed among 4-5 other roles.
- Implement Survey Logic Whenever Possible: Sound survey logic improves a respondent’s experience dramatically. Survey logic allows respondents to only see questions and/or answer choices that are relevant to them, based on how they answered previous questions. There are few things more frustrating in a survey than being forced to answer questions that aren’t relevant. If you need a respondent to remember a previous statement, remind them using survey logic. Human memory is fallible, and a well-placed reminder can reduce the most frustrating of survey outcomes: inconsistent responses.
- Mutually Exclusive and Comprehensive Answer Options: It should never be ambiguous about how to answer a question. All answer choices should be clearly defined, comprehensive in nature, and mutually exclusive. There should always be a relevant answer choice. If there’s no way to be comprehensive, then it’s customary to include an “Other” option where the respondent can fill in their own response.
- Consistent Scales: For surveys that contain multiple rating questions, keep the scales consistent. Each type of scale has its own merit, whether it’s a four-point scale, a seven-point scale, or a 10-point scale (or any other type of scale, for that matter). Whatever you determine to be best for the research objective, keep it consistent throughout the survey. It improves the respondent experience by making it easier to answer questions. Well-designed, easy-to-answer questions will produce good data.
- Simple Language: Keep it simple! It works in many other areas of life and works in surveys, too. Remember that your population will define what “simple” language means. Speak the language of your specific audience and they will thank you by providing good data – and by not dropping out of your survey.
- Take Advantage of Varied Question Types: Single-select, multi-select, rating, ranking, grid, dropdowns – the list goes on. There are many ways to ask a question but knowing the right application of each is the differentiating factor. The right question type is easy to interpret and provides data that is easy to analyze. Remember to limit open-ended questions. Nothing fatigues a respondent faster than too many essay-style responses. You can get good data from two or three open-ended questions, but more than that and you will see a decline in quality and/or higher dropout rates.
- Be Thoughtful About the Number of Questions You Ask: Your survey should ask the least number of questions possible that will still allow you to gain meaningful insight into your research objective. That number may be five questions, it may be 35 questions. Different populations have different tolerances for the length of a survey. Know what that tolerance is and adhere to it. And perhaps most importantly, don’t ask redundant or irrelevant questions.
If you follow these survey design tenets, your surveys are more likely to deliver quality data that translates into actionable results.