October 14, 2016
How to Create an Online Survey
Online surveys are increasingly popular—and there’s no doubt why. These simple research instruments can be used to measure everything from customer satisfaction to brand awareness. Plus, in the digital era, conducting survey research is as easy as 1, 2, 3:
- Upload survey questions to an online survey tool (e.g. SurveyMonkey or Typeform)
- Distribute the survey via email, social media, and/or SMS
- Get virtually instant results You can even create online surveys from your mobile device. But they’re also problematic: Online surveys often have low response rates, unrepresentative samples, and biased results. Respondents frequently abandon surveys that are too long and skip over items they don’t care to answer. As a result, online surveys have serious data quality issues. By learning some online survey design best practices, you can improve your response rates, enhance the quality of your data, and gain powerful insights into your survey population. Follow these 5 steps to create an effective online survey:
Step 1. Define your research question
First, ask yourself what question you are trying to answer. Doing so will help you understand the purpose of your research. Here are a few sample questions:
- What do Fortune 500 CMOs plan to prioritize in 2017?
- How satisfied are our customers with the service we provide?
- What percentage of small business budgets are dedicated to digital marketing?
When deciding on a research question, be mindful of scope. Questions that are too narrow don’t reveal enough insight—limiting the value of your investment. Questions that are too broad result in lengthy surveys that discourage respondents from participating.
Step 2. Determine your sample size
Before you can determine your sample size, you’ll need to establish:
- The size of your population
- The margin of error you’ll use
- The confidence level you’ll use
First, determine the size of the population you’re trying to survey. If you’re conducting a customer satisfaction survey, your entire customer base is your population. If you want to know what Fortune 500 CMOs plan to focus on in 2017, your population is the total number of Fortune 500 CMOs in existence. Next, choose your margin of error. This percentage impacts the accuracy of your survey results. It’s best to choose a margin of error between 1% and 10%. (Hint: It’s standard practice to use a 5% margin of error.) Here’s how it works: Your customer satisfaction survey reveals that 70% of your customers are happy. With a 5% margin of error, you can be confident that 65-75% of your customers are happy. But with a 10% margin of error, you can only be confident that 60-80% of your customers are happy. Once you’ve chosen your margin of error, it’s time to pick your confidence level. This figure determines the reliability of your sample. A high confidence level means a separate sample is likely to reveal the same result. A low confidence level means a separate sample is unlikely to reveal the same result. You can choose a 90%, 95%, or 99% confidence level. (Hint: The most common is 95%.) Finally, determine the size of your sample. To avoid human error, I recommend using SurveyMonkey’s sample size calculator. As you can see, if you have 5000 customers, you’ll need roughly 360 survey respondents for accurate results.
Step 3. Design your survey
As a marketer, you should always strive to reduce your customer’s effort. Online surveys are no different. When designing your survey, follow these guidelines:
Be aware of respondent fatigue
Because online survey research requires little effort, people get asked to take part in surveys weekly, if not daily. The result? Respondent fatigue. Respondent fatigue sounds complicated, but it’s not. It just means that people are tired of taking surveys. As a result, they ignore your requests (leading to lower response rates) and drop out of lengthier surveys (resulting in higher abandon rates).
Keep it short and simple
An online survey should never take longer than 15 minutes to complete, suggests Qualtrics. But you can boost response rates and reduce abandon rates by designing a survey that takes less than five minutes to complete. In that time, most respondents can answer about 10 questions. If your research question is complex, you may end up with a longer survey. That’s okay. In fact, SurveyMonkey suggests that completion rates don’t drop significantly until the 7- or 8-minute mark. And because respondents spend less time completing longer surveys, that’s usually enough time to answer 15-20 questions.
Design mobile-friendly surveys
About 20% of respondents complete surveys from their mobile devices, according to Greenbook’s 2014 GRIT Consumer Participation in Research Report. To prevent mobile respondents from abandoning your survey, choose a provider that offers responsive designs.
Use fast-to-slow progress bars
Progress bars can encourage or discourage respondents from completing an online survey. If little progress has been made after considerable effort, the respondent may abandon the survey. But if significant progress is made quickly, the respondent is likely to complete the survey. Constant progress bars, which accurately depict respondents’ progress, have little to no impact on survey completion rates. But fast-to-slow progress bars decrease abandon rates by 20%, according to research scientists from Google, Gallup, and City University London. Fast-to-slow progress bars advance quickly at the beginning of the survey (implying the respondent has made more progress than they actually have), but slow down dramatically near the end of the survey (to compensate). Respondents are thus less likely to jump ship, because they feel they’ve completed the majority of the survey. But be careful: If your customers realize they’ve been misled, they may feel alienated.
Step 4. Write your survey questions
Now it’s time to create questions for your survey. There are 2 types of questions to choose from:
1. Closed-ended questions
Closed-ended questions force respondents to choose specific answers. They’re often preferable to open-ended questions, because they’re much easier to analyze. Multiple choice questions are closed-ended questions. These questions require respondents to select one or more options from a list. Rating scales are also closed-ended questions. They assess respondents’ thoughts, feelings, and beliefs along a continuum. The most widely used rating scale is the Likert Scale (shown above). Scholars claim it’s the most reliable way to predict behavior based on self-reported attitudes. Pro Tip: If you use rating scales, keep the format consistent throughout the survey. Say your first question assesses customer satisfaction using a 5-point scale, with “extremely satisfied” on the right and “extremely unsatisfied” on the left. If the next question assesses customer loyalty along a continuum, continue to use a 5-point scale, place “extremely loyal” on the right, and place “extremely disloyal” on the left.
2. Open-ended questions
Open-ended questions allow respondents to respond freely. While they’re not as easy to analyze as closed-ended questions, content analysis software can be used to analyze the frequency of certain words or phrases. In online surveys, comment boxes are used to collect answers to open-ended questions. Aim to write questions that are easy to understand. To prevent confusion while increasing the quality of your survey data, follow these guidelines: **Never lead the respondent **Incorrect: Wouldn’t you like to wake up at 9:30 a.m.? Correct: What time do you wake up in the morning? **Never use jargon or slang **Incorrect: How often do you reach out to your sales rep? Correct: How often do you contact your sales representative? **Always use plain language **Incorrect: How frequently do you purchase bottled water? Correct: How often do you buy bottled water? **Never use the word “and” in a question **Incorrect: Was the presentation informative and engaging? Correct: Was the presentation informative? **Never ask more than one question per survey question **Incorrect: How helpful was the customer service representative? Was he/she friendly? Correct: How helpful was the customer service representative? Once you’ve developed a list of questions, assess each question. Ask yourself: Will this question reveal insight into my research question? Continue to revise your questions until you feel confident.
Step 5. Set up (and send) your online survey
Congratulations! You’ve finished preparing your survey. Now all you need to do is set it up online using the provider of your choice, upload a list of potential respondents, and…hit send! What tips have helped you design more effective online surveys? Let us know in the comments.