Designing Marketing Surveys - dummies

By David Semmelroth

Even if you have the most extensive and accurate customer database in the world, it can’t tell you everything you might want to know about your customers. To answer some questions, you need to actually ask your customers, and you generally do that in a survey.

You can conduct marketing surveys in a variety of ways. You can interview people face to face. You can call them on the phone. You can distribute surveys by mail or e-mail. Offering visitors the opportunity to conduct a survey is now commonplace for websites. These methods vary in both cost and effectiveness. That’s why it’s important to understand that survey research is a science unto itself and requires some professional expertise to make it effective. This article points out a couple key considerations that are fundamental to a successful survey research project.

Think carefully about the audience for your survey

The first consideration is to identify whom you want to survey. Understanding your audience is critical to interpreting your survey results. For example, a survey may tell you that you have low brand awareness for your mobile app. When interpreting such information, it’s pretty important to understand whether the survey responders even own mobile devices.

Another consideration is the size of your audience. You’ll be using statistical techniques to analyze your survey results. If you don’t get enough responses, these techniques won’t yield meaningful results. The minimum number of responses required varies, depending on what sort of analysis you’re doing. But a basic rule of thumb is that fewer than 30 responses doesn’t yield anything meaningful. Often, several hundred or more responses are needed.

Remember, the number of responses determines whether results are meaningful, not the number of surveys. If you’re conducting an e-mail survey, for example, you need to account for non-responders. Many of your e-mails will never be opened. Many more will be read and discarded with no response. You may need to send out 10, 20, or even 50 e-mails to generate a single response.

Think carefully about your survey questions

Measure twice, cut once. The adage applies to survey research as well as to carpentry. An effective survey requires thoughtful design. The length of your survey and the way you phrase and order your questions can dramatically affect the quality of the results.

If a survey is too long, people lose interest. They’ll abandon the survey without completing it. Worse, they’ll complete it, but mindlessly. In this latter case, the accuracy of the results degrades according to how far down the survey a particular question was placed.

The order of the questions is also important. You don’t want the answer to one question to influence the answer to another. For example, you may ask someone what’s their favorite car of all the cars they’ve ever owned. If you then ask them to rank the quality of several automobile manufacturers, the answer will be biased by the previous question.

Maybe the biggest pitfall in survey research is asking leading questions or otherwise biasing the responses through the wording of questions. For example, a survey that asks responders to state their preference for candidate Joey Smith, Jr. versus Doctor Marcus Jones, JD, PhD, introduces a clear bias toward the experience and education of the latter candidate. That’s an extreme example. Bias can be much subtler. You should review each question with vigilance to make sure it’s objective.

To get the most from your survey research projects, engaging some experts is a good idea when you design your survey. Survey design and statistical analysis are sciences unto themselves. If you’re going to spend the money to learn about your customers through research, you may as well do it right.