How to create a survey 

Surveys are great tools for collecting large amounts of quantifiable data about how a user or customer feels about something. However, it is important to make sure your survey is well-thought-out and structured to yield reliable results. The following guide will help you plan and structure your survey to obtain valuable insights. 

What you’ll need: 

• 60-90 minutes to create, 2-4 weeks to run the survey. 
• Survey software, see below. 

• Pens, notebook, laptop, or your preferred notetaking method. 
• A friend to review your progress. 

Steps: 

Define a clear and attainable goal for your survey. 

Ask yourself: ‘What is the question that I need to answer from my survey?’ Make sure that your goal is specific and measurable. For example, “I want to understand people’s first choice” is a vague question. “I want to understand what are the key factors that affect what food people choose to buy and how much cost influences their decisions” is a clear and measurable question. Remember, surveys are useful for gathering quantifiable data, so your survey should try and identify ‘how much’ or ‘how many’. 

List out all the questions you want to answer. 

Get all the questions you want to answer in one list. This will help you prioritise and organise them appropriately. When you’re listing all your questions, think about asking closed questions rather than open questions. Having multiple-choice questions or number scale questions are quicker to fill in and easy to quantitatively analyse. 

Now, organise your questions with the easiest questions at the top of the survey. 

You should put the easiest questions at the top of the survey. It is also important to prioritise your questions and only ask what is necessary. Surveys that are too long can lead to survey fatigue, which can lead to low response rates or incomplete submissions, which affect the validity of your survey results. While there is no ‘right’ number, you should aim for around 10-12 questions maximum. 

Make sure that you’re only asking one question each time. 

Follow the rule of one question à one answer. A question like: “What are your thoughts on vegan food and the price of food currently?” is bad as they’re asking two questions at the same time. Your participant might have different thoughts on each question. 

Place a clause at the top explaining what the survey is about. 

This is your opportunity to inform the respondents of the reasons why you collect their personal data, where you store it, for how long you process it, and in which ways you will use it in the future. This should go at the top of the survey. For more information, see this guide on protecting users’ data. 

Get your friend to review your survey before sending it. 

Make sure you have someone look over your survey and proofread it before you send it. This will help you catch any areas that are unclear to your participants. As you’ll have been working on your survey for a long time, there are some things you might miss that a friend will pick up very quickly. 

Best Practices: 

Below are some best practices that you should follow when creating a survey. This will ensure your survey makes sense and yields reliable responses. 

Don’t ask leading questions. 

When you’re writing your questions, make sure you’re not putting your opinion in the question. Your questions should be balanced and neutral. For example, instead of asking “How would you rate your delicious meal today?”, it is better to ask, “How would you rate your meal today?” 

Avoid using absolutes, emotive, or evocative language. 

Similar to asking leading questions, make sure that you’re not asking questions that share an opinion or prevent nuanced answers. For example, “Do you always eat breakfast?” has an absolute with always. Instead, you should ask “On average, how many times do you eat breakfast weekly?” Similarly, asking a question such as “What is your opinion on the failings of current government policy?”, features emotive language, and we should remove ‘failings’ from our question. 

Make sure that you’re being specific. 

Where possible, make sure you are being specific and clear, as someone might interpret your question differently. For example, rather than asking “What is your income?”, you should ask “What was your total household income before taxes from April 2022 to March 2023?” When you’re using ordinal scales (on a scale between 1-10, etc.), make sure you label the extremes (1 being poor and 10 being excellent). 

Avoid technical terms, jargon, and internal terminology. 

Make sure that you’re using accessible language that is easy to understand. If you’re using acronyms, make sure you explain what they stand for. Never use a long word when a short one will do. 

Use Likert Scales to understand how much someone agrees or disagrees. 

There are examples online, but these scales usually have 5-7 responses, with a middle neutral category. For example, a 5-point Likert Scale would include: Strongly Disagree; Disagree; Neither Agree nor Disagree; Agree; Strongly Agree. 

Allow all possibilities. 

When appropriate, you should make sure that respondents can answer as thoroughly as possible. For example, if you have an employment list and a participant is both a student and employed, you should allow them to answer both. Furthermore, if the categories you list don’t list all possible choices, you should include an ‘other category’ that allows participants to write in their responses. 

Only ask for the information you actually need. 

There is no such thing as ‘nice to have’ data - make sure that you’re only asking for what you need. Remember, asking for too much can lead to survey fatigue or can make people reluctant to fill a survey in thoroughly. If you’re collecting identifying data (gender, sexuality, race, ethnicity, etc), make sure you are asking this at the end and being compliant with your data. 

 

Further Support 

For more information on developing your entrepreneurial skills or launching your business idea, book a 1:1 here.