Ask about overall satisfaction, how easy the experience was, whether the issue was resolved (for support contacts), whether expectations were met, and how likely the person is to recommend you. Finish with one or two open questions. Five to ten questions is usually enough, and two to four is often better right after a support chat or call. The rest of this guide explains how to choose and word them.
Start with the one thing you want to learn
Before you write any question, decide what the survey is for. A survey after a support call measures something different from a survey about a purchase or a product trial. Pick the one or two decisions the answers should inform, such as whether to change the returns process or whether the new onboarding works. Each question should earn its place by helping with one of those decisions.
A core set of questions
These questions cover the most common goals. Choose the ones that fit your situation, and keep the set short.
- Overall satisfaction. “How satisfied were you with your experience today?” (Very dissatisfied to Very satisfied)
- Likelihood to recommend. “How likely are you to recommend us to a friend or colleague?” (0 to 10). This is the question behind Net Promoter Score (NPS).
- Effort. “How easy was it to get the help or information you needed?” (Very difficult to Very easy). This is the basis of a Customer Effort Score.
- Resolution. “Was your issue resolved?” (Yes, No, or Partly). Use this for support contacts.
- Expectations. “Did our product or service meet your expectations?” (Exceeded, Met, Fell short)
- Value. “How would you rate the value for money of our product or service?” (Very poor to Excellent)
- Open feedback. “What is one thing we could do to improve your experience?”
For a support contact, you might also ask how satisfied the person was with the representative’s help, or how quickly you responded. Keep those in the set only if the team can act on them.
What CSAT questions are
CSAT stands for customer satisfaction. A CSAT question asks people to rate how satisfied they were with a specific interaction or experience, usually on a short scale. “How satisfied were you with your experience today?” is a typical CSAT question. Because it asks about one moment, it works best when it is sent soon after that moment.
Wording rules
Clear wording matters more than the number of questions. These rules prevent the most common problems.
Keep one objective per question
Watch for the word “and.” “Was customer service helpful, and would you buy again?” asks about two things. Split it into two questions, or keep only the one you need. A question like “Was your problem resolved?” is clearer than one that asks about both the support and the outcome at once.
Do not assume the customer’s experience
An assumptive question takes something for granted. “How do you like our new loyalty card?” assumes the person has one. Ask first: “Have you used our loyalty card?” Then follow up only with those who answered yes. People who feel the survey does not understand them often stop answering.
Remove leading qualifiers
Do not add reasons for the answer you want. “We recently spent a lot of money on this website. Do you like it?” pushes people toward a yes. Ask “Did you like our website?” instead. In general, take words out of the question rather than adding them.
Check grammar and double negatives
Double negatives are easy to miss in drafts. “Was our service not unhelpful?” is hard to answer. Ask “How helpful was our service?” instead. Proofread the survey, and ask a colleague to read it, because a sloppy question can make a respondent doubt the survey is genuine.
Avoid near-duplicate questions, but repeat on purpose if needed
Asking “How satisfied were you?” and then, five questions later, “I was happy with the service,” annoys people and adds little. Remove duplicates. There is one exception. If you suspect people are agreeing with everything, a deliberately repeated question with different wording can reveal it. If many people give contradictory answers to the pair, review the original wording or treat those responses with caution. The Likert scale guide explains how to balance scales.
Put open questions at the end
Open questions ask people to explain in their own words. They produce useful detail, but they take effort. Place them after the rating questions, so that people have answered the core questions before they reach the harder ones. If you put an open question first, many people will see the survey as work and leave.
When to send the survey
Send the survey while the experience is fresh. A short survey after a purchase, a support ticket, or an onboarding step gives more accurate answers than a long one sent weeks later. Avoid sending the same person many surveys in a short period, because repeated requests can lower response rates and sour the relationship.
Test before you send it
Read the survey on a phone, then ask two or three customers to answer it while they talk through their thinking. Check where they hesitate and which questions they skip. The pilot testing guide covers this process in more detail.
Common mistakes
- Asking too many questions. A long survey loses people before the most important questions.
- Mixing two objectives in one question. The answers cannot tell you which part caused the result.
- Asking about a product the customer has not used. Filter first.
- Using leading wording. Positive answers that come from wording are not evidence of satisfaction.
- Putting the open question first. It signals extra work and can lower completion.
For question banks organized by sales, B2B, and service contexts, see the customer satisfaction survey questions guide. For the wider design of a research project, start with the market research survey questions guide.