I was recently tasked with writing a survey for a client as part of the UX research process, as a precursor to user interviews. Surveys can be a great way to gather data around your users, who they are and what they do. User interviews however, focus more on the how and the why - the personal experiences of each user. Therefore surveys are quantitative, and their output might give you some nice numbers and graphs. User interviews are qualitative, and outputs could be personas, which help the designer to consider how the user feels.

Using a survey not only gives you quantitative data around your users, but also around the problem space of your project. Will you find out that your users don’t think that your problem is valid? Will they in fact identify a different problem that you have not thought about? Maybe! Surveys are a simple way to gather such insights.

Let’s be honest, there probably aren’t many people who love filling in surveys, so writing a good one is important. Survey abandonment is common. When gathering quantitative data in a survey format, there are many things to consider to ensure a high response rate, and that your repondents have a positive experience. But what are they? Here I will share what I have learnt during my own survey writing experience.

Preparation is important, in order to write a survey that is applicable to the audience that you are targetting. If one wanted to survey respondents in the boat building profession, it would help to have some understanding of the boat building process, what jargon is used around the profession, and what the day to day might look like. However, when conducting this research, remember not to assume that all boatbuilders use the same process, same jargon and have the same daily routine. Research should inform your survey questions, but not rigidly define them. All I am saying is, a quick whizz through the boat building wikipedia page could give you some idea of what is appropriate. At the time of writing this article, the page begins with the sentence, “Not to be confused with shipbuilding”. How about that!

If you are working for clients to create a survey, it also helps to get their knowledge and feedback on the potential problem and users. After all, they will definitely have some ideas about it. Maybe they themselves identified this problem as pain point in their own lives. Asking your clients about what sort of questions they want to ask users around the problem will be not only helpful to you, but also ensures that you client is invested in your work.

Carefully consider the language that you use. It is generally best to keep language simple so that respondents easily understand your questions and do not feel alienated. Avoid jargon unless it is directly and commonly related to the professional or cultural group that your respondents belong to. Keep language formal so that the tone is professional and so that respondents feel that their answers are being used responsibly and legitimately. Phishing, where scammers and internet rascals seek to gather information from people for devious and criminal purposes, is on the rise. Therefore, securing trust by using proper spelling, grammar and formalities is increasingly important.

Another way to create trust in your survey is to introduce it well with a wee blurb about its purpose, and assuring respondents that the data gathered is not public and to be used for this project only. Provide email addresses or contact numbers for anyone with questions, and sign off your introduction with your name. If people can look you up and see that you are a real person, who is in fact a UX researcher or developer, they will have greater peace of mind.

However, also ensure that your explanation of the purpose of the survey does not lead respondents to answer in a certain way. Your users are experts in the problem, not experts in the solution. If you explicitly say what you are trying to research, or in any way suggest a solution, data from your survey can be skewed.

Is your survey long? Chances are that it will be abandoned. Keeping surveys short and sweet increases chances of a high response rate. SurveyMonkey gathered data around survey participation on their platform, and discovered that “surveys that took more than 7-8 minutes to complete” saw “completion rates dropping anywhere from 5% to 20%”. They also found that if people did perservere with arduous surveys, the longer the survey the shorter amount of time spent on each question. This could lead to inaccurate response data, which of course is not desirable.

It is not just the amount of questions that make a survey long however, but also the difficulty associated in answering each question. If a question is wordy, vague or unclear then it becomes harder to respond to. When the answer to the question is, “it depends”, then how does one answer succinctly? Keep questions clearly defined, with yes/no, multichoice or scaled answers. Where applicable, give an option for respondents to select an option of “other” and specify their answer. This gives allowance for if you have made incorrect assumptions of what the answer to your question might be.

Free text, or essay, questions also lengthen a survey considerably. Remember that the survey is for quantitative, not qualitative, data. Therefore keep free text questions to a minimum, and if they are really necessary ensure they are short-text. Most free text questions are better asked at the user interview stage. This is because, within the survey, free text answers lack the context that makes them meaningful and easy to understand, as well as synthesize. In general making people think too much, and encouraging essay writing, is not going to make your survey experience enjoyable.

Ok, so now your survey is not long and arduous, but does your respondent know that? If your survey platform offers the feature, use a progress bar to show how far respondents have come, and how far they have to go. Otherwise, try to have someone do a dry run and note approximately how long the survey takes in your introduction. You can also use section comments or descriptions to let your user know where they are in the land of your survey. For instance, a little note like, “Section 1 of 6”, or, “Section 3 of 6 - halfway there!” will be appreciated.

Ensure that your questions to not create skewed or inaccurate answers. If your questions are leading, then respondents may feel encouraged to answer in a certain way. If it is difficult to avoid language in your question that could be considered leading, then use a likert scale to collect answers. A likert scale is a scale of options where at each end will be the extreme. For instance, a likert scale that is commonly used is one that determines respondent agreeance with a statement. On one end of the scale is the label, “Strongly agree”, and on the other “Strongly disagree”.

Ensure that your scale is equally weighted if labelling every point on the scale. The middle option, in this instance, should be something like, “Neither agree nor disagree”. If points on the scale read something like, “Strongly disagree”, “Very much disagree”, “Somewhat disagree”, “Hardly disagree”, “Agree”, you can see how answers will be skewed towards disagreeance. Likert scales are also great for some of those “It depends” questions, where there is not strictly a “Yes” or “No” answer but maybe a range due to differing circumstances surrounding the question.

Use a survey making service that allows you to branch questions. Question branching allows you to expose your users to certain questions or avoid them, depending on how they answered preceding questions. If you ask the question, “Are you interesting in becoming a boatbuilder?”, and the respondent selects, “No”, then it doesn’t make much sense to ask the question, “What institution would you like to go to learn more about boat building?” afterwards. This will make the survey potentially shorter for some respondents (win!) and also decrease annoyance at being asked seemingly pointless questions.

Want another excellent reason to lower the chance of respondent annoyance and consider all of the above? Happy respondents might agree to be interviewed in the qualitative data gathering phase that is user interviews! The end of your survey is an excellent time to ask if respondents are interested in being interviewed more in depth about their experiences surrounding the issue that you have identified with your survey. Here you can collect contact details, perhaps email addresses, to arrange an interview session.

Overall, ensure that you do everything you can to make your survey quick and easy to participate in. Not only will you receive more resonses, but your data will be more accurate and you may even have the chance to gather more information from respondents at a later date. If survey respondents have a positive experience overall they may also have positive future associations of your product - at the very least their view of your product and organisation will not be tarnished by a tiresome or difficult process.