How To Write A Good Survey
March 26, 2012 Leave a comment
I recently had the opportunity to look over the results of a participant survey for a organization that put on a week long conference. While I did not write the survey, I was analyzing the data to help them out for next years conference. While I was looking at the results, I remembered that there were some good rules of thumb to remember when writing a survey to get better responses and response rates.
- Purpose of survey: Before you start, you need to identify the purpose and what are the “key learnings” you want to know. List them out. The questions you will ask for a market depth survey is very different than a customer satisfaction survey. Be sure to include demographics to make sure you get your buyer profile.
- Execution of survey: How the survey is going to be executed is just as important as the survey itself. Included in this is the list you are sending the survey. Mail surveys versus email surveys have different Expected response rates. A cold list versus an online panel versus a customer list will get varying responses.
- Length & Complexity: This needs to be considered in conjunction with who you are sending the survey to. Customers are more likely to answer longer and more complex surveys. Those with no connection to your company, are not as likely to answer that same survey. Also, you need to consider any type of incentive to promote your response rate. The longer and more complex, the better the incentive.
- Question bias: This is something even seasoned survey writers fall in to. I remember a marketing class I had that had a handout showing questions from statistically accurate surveys. On the surveys, there were essentially the same question, but with dramatically different results because of the wording of the questions.
- Ratings vs Rankings: Ask five different market researchers the scale system they like, and you will likely get six different answers. Odd scale versus even scale. Five-point versus seven-point. Six-point versus ten-point. These are all things to consider. The more points the more finite between the values. Odd scales have a mid-point, when even scales force someone to be either positive or negative. Asking rankings are generally not a good idea. This is because of two main reasons. You cannot synthesize these answers across respondents. My difference between #1 and #2 is likely different than your difference between #1 and #2. Also, my difference between #1 and #2 could be different between #11 and #12.
- Open-Ended Questions: When filled out on surveys, these are great. However, it is only when they are filled out. A lot of people tend to skip these questions. Be sure to limit them and use them as only as needed. Also, be sure to use some form of text analytics if you have a larger group of respondents.
Surveys can help shape strategy and answer pressing operational and product questions. However, be sure to make sure that the survey serves your needs based on posing good questions.