Most customer satisfaction surveys suffer from uneven customer representation, which leads to inaccurate data. As the saying goes, “garbage in, garbage out.” And let’s face it, garbage should never be the basis of your business decisions. You deserve facts, not fiction. So here are a few ways to improve your customer survey sampling methods to achieve balanced customer representation and high quality data.
Customer Survey Sampling Error #1: Non-Response Bias, Some Don’t Respond
Just because you openly invite all customers to take your survey does not mean you have a random sample. The difference between an open invite and randomized responses is critical because satisfied customers with time to spare tend to be the population most apt to take your survey.
Similarly, you are more likely to hear from repeat customers than from one-time customers—yet customers who aren't coming back may have the most useful feedback. Here are a few ways to get a representative customer survey sample:
• Start by dividing your customer database into categories, such as: current most-profitable customers, current less-profitable customers, lapsed customers, long-gone customers, prior most-profitable customers, etc.
• For each customer category, determine how many customers you need to survey to achieve a reasonable confidence interval and error rate. Then randomly select customers from each category you want to test.
• Proactively ask the selected customers for their feedback. You may need to offer incentives. Test different types of incentives, but make sure they’re not related to the products or services you sell, because this introduces yet another type of bias—those who like your product will be more inclined to take your customer survey.
• When you invite customers to take your survey, ask if they would prefer to share their thoughts by phone at a convenient time. Creative, busy, and dissatisfied customers may not be receptive to multiple choice questions—yet for accurate, actionable data, you need to hear from your entire population. Again, using an incentive may need to be part of your customer survey strategy.
Customer Survey Sampling Error #2: The Medium Controls the Audience
Do you offer your survey online? By IVR? Mail? Email? How you provide your customer survey has a huge impact on who takes it. Here are a few ways to increase responses from all your survey channels and correct for channel-related distortion:
• Most likely, your customer satisfaction survey is online, which provides for a variety of delivery options. Set up your survey so customers can dial an 800 number, visit a URL, or use their mobile to text their answers. The more options you offer, the more respondents you’ll get.
• Be willing to receive survey responses the way customers want to give them. For example, if you email your customers with a link to your survey, tell customers that if they prefer, they can simply reply to the email with their thoughts about your company and their experiences—they don’t have to take the survey. Then, use a scientific coding technique to incorporate their open-ended comments with the rest of your text analysis.
• Finally, tell customers the survey is short, and then keep good on the promise.
Most customer surveys are plagued with biases—and sampling errors are just one type. But with strategies to prevent sample bias and channel distortion, you’ll be one step closer to accurate data, leading to decisions based on fact, not fiction.
Gain a new perspective. Start for free.
Interaction Metrics is a customer experience agency that maximizes the value of experience planning, satisfaction surveys, mystery shopping, customer interviews, and customer service evaluations. Only Interaction Metrics Findings Reports combine actionable customer experience metrics with specific recommendations for how to improve.