There are many places you can go wrong in constructing and delivering a customer satisfaction survey that can keep you from getting an accurate picture of your customer experience. And most companies do go wrong in their survey design and delivery–often in multiple ways. (Alarming but true: If you send out a defective survey, it might be worse than not surveying your customers in the first place, because of the risk that the data you get back will be invalid but nonetheless used to guide company strategy.)
What follow are 13 principles of correctly surveying your customers on their experiences with your company, best practices that are both scientifically derived and informed by my own work as a customer service and customer experience consultant and designer. (I know it’s an odd number, but all of these principles are essential, and #13 perhaps most of all.)
1. Ask for the customer’s overall rating first. You don't want to influence this answer by asking smaller, more nitpicky questions before you get to the biggie; asking your customer several individual questions and onlythen getting around to asking for an overall rating reduces the validity of that all-important rating.
2. Limit your survey to a reasonable number of questions. You want your responses to come from the mainstream of the people you’ve sent the survey to, rather the response be skewed to only recipients with the time and patience to slog all the way through a too-long survey. (Do you really only want to know the preferences of customers who took the time to answer a thirty-question survey without leaving a single answer blank?)
3. Only offer a small number of ratings choices. Generally, I suggest 1-5 as your rating scale; 1-3 might be even better. Definitely don’t go as high as 1-10; it’s fruitless to expect a survey participant to choose between 10 different possible ratings–c’mon! What does a “6” even mean?)
4. Phrase your response categories in the most concrete language possible. Don’t use rating categories like ‘’excellent’’; ‘‘excellence’’ is essentially indefinable, so look for something that is based on your customer’s own experience. Consider calling your top rating something emotive and simple, such as ‘‘Loved it!’’ or “Wonderful!”
5. Don’t expect your customers to be mathematicians. Demanding participants make calculations along the lines of, ‘‘estimate your chances of returning to our store this month in terms of percentage of 100’’ will create confusion and frustration.
6. Don’t ask intrusive demographic questions such as income, gender, or age without making the responses optional. Don’t assume that respondents will trust your privacy practices. (Would you?)
7. Don’t use internal jargon. You need to speak the language of your customers, rather than your internal lingo. (However: if all of your customers are from the same industry, as is common in B2B, you can certainly use jargon that is familiar to them.)
8. Surveys are most meaningful if completed soon after a customer’s experience. This means that you need to survey customers soon, and that you need to close the window for accepting responses not long after. (An exception to this principle, of course, if you’re sending out a survey asking about a product’s longevity and such.)
9. Don’t hassle recipients for not filling out your survey. Maybe remind them once. I wouldn’t remind them twice. These are your customers; they’re not obligated to do what they’re not interested in doing.
10. Include a free-form text field or fields to leave room for novel responses that you may not have even considered and to offer your customers an opportunity to express themselves.
11. Be sure to respond personally and promptly after receiving strongly negative feedback. And don’t set a batch of surveys aside for later en masse response without scanning them in a more timely manner for negative responses that require immediate replies.
12. Be sure to thank anyone who offers personal praise on a survey. A handwritten note is a wonderful way to accomplish this if you have the customer’s physical address. An email is also fine, as long as it’s clearly from a real person and not boilerplate.
13. If you send out similar surveys over time and expect to compare results, it’s essential to understand that you cannot change anything in your delivery approach, introductory materials, or survey content without making your results impossible to compare as apples to apples. One of the consistent findings of social psychology and behavioral economics (aka “psychology with a name-change for marketing purposes”) is the often-intense and disproportionate effect of what would seem to be small, even trivial, changes in circumstances. In the world of surveys, this means that, for example, there can potentially be huge effects from changes as apparently minor as:
- Changing your survey’s introductory language, or even its subject line
- Changing whether your survey’s delivered by email or weblink
- Changing the number of days until it’s sent out and how long it remains open for responses
- Changing the number of choices per question
So what to do if you absolutely have to change an aspect of your survey? You’re on your own, here buddy. At the very least, make a careful internal note of which element you’ve changed and caveat your results internally to avoid putting too much reliance on cross-survey results.
[A note about this article's content: This is the most comprehensive peek under the lid I’ve ever offered on great survey design. However, portions of this have been recycled, with updates as appropriate, from a related article of mine published here in 2014.]
