The best customer satisfaction survey example is one that matches the experience you want to understand. A question about a recent support conversation cannot tell you everything about a customer’s relationship with your company. A loyalty question cannot explain which checkout step failed.
Below are ten survey approaches with questions you can adapt. The examples are original illustrative templates, not surveys used by named companies or validated replacements for proprietary instruments. Choose a focused set, test it with the intended audience and define how you will act on the answers.
Choose a survey by the decision it supports
| Approach | Useful question | Typical reporting |
|---|---|---|
| NPS | Would customers recommend the company or product? | NPS and reasons. |
| CSAT | How satisfied were customers with a defined experience? | Rating distribution and favorable share. |
| CES | How easy was a particular task? | Defined ease score and obstacles. |
| Service-quality expectations | Where does perceived service fall short of expectations? | Paired item gaps, where appropriate. |
| CX pulse | Did an experience work, feel easy and leave a positive impression? | Separate dimension ratings. |
| Post-purchase or post-interaction | What happened at this journey stage? | Stage-specific answers. |
| Customer health input | Is the account making progress toward its goals? | Contextual answers alongside other evidence. |
| Voice of customer | What needs and friction should we investigate? | Themes within a wider listening program. |
| Relationship and loyalty | How does the ongoing relationship feel? | Relationship NPS, intentions and reasons. |
| In-app or website feedback | Did someone complete the task on this page or flow? | Completion answers and reported barriers. |
These categories overlap. A post-interaction survey can contain CSAT or CES; a VoC program can include relationship surveys. They are not ten interchangeable scores.
1. Net Promoter Score survey example
Use NPS when recommendation is relevant to the relationship you want to measure. Specify whether the question concerns your company or a particular product. Keep the scope consistent across reporting periods.
How likely are you to recommend [company/product] to a friend or colleague? 0–10, from not at all likely to extremely likely. What is the main reason for your answer? Optional free text.
Bain’s NPS scoring method classifies 9–10 as promoters, 7–8 as passives and 0–6 as detractors. Calculate the percentage of valid respondents who are promoters minus the percentage who are detractors. Passives remain in the denominator.
For an illustrative 40 valid answers with 20 promoters, 10 passives and 10 detractors, NPS is 50 − 25 = 25 points. It is not a satisfaction percentage. Review reasons and respondent coverage rather than treating the score as proof of future purchases.
2. Customer Satisfaction survey example
CSAT suits a specified experience, such as the support interaction that just ended. State what the respondent is rating so product dissatisfaction does not become an unexplained rating of an individual agent.
How satisfied are you with the support interaction about [issue]? 1 Very dissatisfied 2 Dissatisfied 3 Neither satisfied nor dissatisfied 4 Satisfied 5 Very satisfied What most influenced your answer? Optional free text.
For this five-point example, report ratings 4 and 5 as favorable responses divided by all valid 1–5 answers, multiplied by 100. This follows Qualtrics’ common CSAT calculation. If 18 of 24 valid respondents select 4 or 5, the illustrative result is 75%.
Keep invitation count and response count visible. Exclude unanswered or not-applicable responses from the rating denominator and report them separately. A resolved ticket status does not establish that the customer considers the issue resolved; use a separate outcome question if that distinction matters.
3. Customer Effort Score survey example
CES focuses on the effort involved in a task, such as resolving an issue or completing a purchase. Qualtrics defines CES around a specified task or interaction. Choose one wording and scale direction and retain it when comparing results.
How easy or difficult was it to complete [task] today? 1 Very difficult 2 Difficult 3 Somewhat difficult 4 Neither easy nor difficult 5 Somewhat easy 6 Easy 7 Very easy Not applicable: I did not attempt this task. What, if anything, made the task difficult? Optional free text.
For this original ease-based version, calculate the mean of valid 1–7 ratings and report the distribution; higher means easier. Do not reverse the interpretation because another questionnaire uses “effort” with high values meaning harder. Record the wording, scale, denominator and time window next to the result.
If the task remains unfinished, preserve that context. Low effort to submit a request is different from low effort to obtain a working resolution.
4. SERVQUAL-inspired service-quality questions
SERVQUAL examines expectations and perceptions of service quality. The original 1988 paper describes a 22-item instrument. Its familiar dimensions are reliability, assurance, tangibles, empathy and responsiveness; the developers’ earlier working paper describes these five dimensions.
The following original prompts are dimension-inspired examples, not an official short SERVQUAL questionnaire. For each dimension, ask an expectation item and a perception item using the same scale. A reliability pair could be:
Expectation: A service provider should carry out the commitments it makes. Perception: [Provider] carried out the commitments it made in my experience. Rate each separately: 1 Strongly disagree to 7 Strongly agree. Not applicable / Cannot assess are separate unscored options.
Other dimensions can be explored with equivalent paired wording. These original statements describe the perception side:
- Reliability: “The service provider carries out the commitments it makes.”
- Assurance: “The staff explain the service in a way that gives me confidence.”
- Tangibles: “The service materials are presented clearly.”
- Empathy: “The staff take my particular circumstances into account.”
- Responsiveness: “The service provider responds when help is needed.”
These wording adaptations require testing in your setting. Where paired answers are appropriate, a local item gap can be defined as perceived rating minus expected rating for the same respondent. With illustrative ratings of 4 perceived and 6 expected, the gap is −2. Exclude incomplete pairs and report how many remain. Do not present a custom shortened or weighted questionnaire as having the original instrument’s validation or benchmark comparability.
5. A customer experience pulse inspired by CX Index dimensions
Forrester’s CX Index measures effectiveness, ease and emotion using a proprietary methodology. A simple internal pulse can explore those dimensions, but it does not produce Forrester’s official index.
Did this experience help you accomplish what you needed? 1 Not at all; 2 Slightly; 3 Moderately; 4 Mostly; 5 Completely. How easy or difficult was this experience? 1 Very difficult; 2 Difficult; 3 Neither; 4 Easy; 5 Very easy. How did you feel about the experience overall? 1 Very negative; 2 Negative; 3 Neither; 4 Positive; 5 Very positive.
Add a not-applicable option where the respondent cannot assess an item. Report each dimension separately, with its valid response count. Do not rescale an average to 0–100 and label it an official CX Index or compare it with Forrester benchmarks. Investigate a gap such as high task completion with low ease rather than hiding it inside a composite.
6. Post-purchase or post-interaction survey example
Match timing to what the customer can assess. Immediately after checkout, ask about placing the order. After delivery, ask about delivery. Product-use questions need enough experience with the product. One invitation should not demand ratings for stages that have not occurred.
Were you able to place the order you intended? Yes / Partly / No. How easy or difficult was placing this order? Very difficult / Difficult / Neither / Easy / Very easy. What, if anything, should we change about ordering? Optional free text.
A later delivery survey could instead ask “Did the order arrive by the date we communicated?” with Yes, No, I do not know and It has not arrived options. Route “not arrived” responses appropriately before asking about the delivered item. Keep checkout ease, delivery performance and product satisfaction as distinct questions.
For a service interaction, identify the interaction and ask whether the customer achieved the intended outcome. A thank-you message can offer a feedback link without implying that only favorable responses are welcome.
7. Customer health survey inputs
A customer health score is usually a configured combination of evidence, not a single standardized satisfaction survey. For example, Gainsight’s documentation describes multiple health measures and configurable scorecards. Survey answers can contribute alongside product activity, support history and account context.
How much progress have you made toward [agreed goal] this period? None / A little / Some / Most / Goal achieved / Cannot assess. What is currently preventing further progress? Optional free text. How confident are you that [product/service] can support this goal? Not at all / Slightly / Moderately / Very / Extremely / Cannot assess.
Record the respondent’s role and the goal being assessed. An administrator and an executive sponsor may see different aspects of the account. Treat “cannot assess” as missing evidence, not a failing score. If answers feed a health model, define weights and validity periods and check its usefulness against later outcomes. Do not assume an untested composite predicts churn.
8. Voice of customer survey example
A VoC survey seeks needs and explanations that inform a wider listening program. It is not a universal “VoC score.” Combine it with interviews and existing feedback, as described in our voice of customer analysis guide.
What were you trying to accomplish when you used [service] most recently? Free text. What, if anything, got in the way? Free text. What one change would make the biggest difference to that experience? Free text. May we contact you to learn more? Yes / No; provide contact details only if yes.
Analyze needs as well as proposed solutions. A request for a dashboard may describe a need to confirm progress, which could have several solutions. Preserve the source and journey context, code comments consistently and avoid claiming that the most frequently mentioned topic represents all customers.
9. Relationship and loyalty survey example
A relationship survey considers the ongoing experience rather than only the latest interaction. Use the recommendation question and 0–10 scale in example 1 with a consistent company or product scope. Add original follow-ups that help explain the relationship:
- “What most influences your decision to continue using [product/service]?” Optional free text.
- “How likely are you to continue using [product/service] over the next [relevant period]?” Very unlikely, Unlikely, Neither likely nor unlikely, Likely, Very likely, Not sure.
- “What would you most like us to improve?” Optional free text.
Keep relationship and event-triggered results separate because their audiences and timing differ. Surveyed intentions are not observed retention. Compare them carefully with defined renewal or repeat-purchase cohorts when appropriate, using customer loyalty measures for that broader evaluation.
10. In-app or website feedback example
Ask about a task or page when the invitation will not prevent the person from completing it. Make dismissal straightforward. A relevant page-level question is usually more interpretable than a generic rating shown during an unrelated task.
Were you able to do what you came here to do? Yes / Partly / No / I was only browsing. If partly or no: What prevented you from completing it? Optional free text. If yes: What, if anything, could work better? Optional free text.
Store the page or workflow context without asking customers to enter sensitive information. Provide a separate support route when someone needs help; a feedback form is not a promise of immediate assistance. Inspect who actually sees the invitation, since people who leave before it appears may be missing from the results.
Make the examples useful before launching
Use plain, neutral wording and test whether people interpret each question as intended. Pew Research Center explains why wording and ordering matter. Ask one concept at a time, provide balanced response choices and avoid making criticism harder to express than praise.
- Define the eligible audience, invitation trigger and reporting period.
- Identify the experience and use a stable scale with clear endpoints.
- Allow relevant not-applicable or uncertain answers and explain their treatment.
- Test the form on the devices your audience uses, including keyboard operation.
- Keep optional comments optional and avoid requesting unnecessary personal data.
- Record invitations, valid responses, missing answers and any changes to wording.
- Give someone ownership of urgent feedback, theme reviews and follow-up.
Start with the questions necessary for one decision. Share results with counts and context, investigate the explanations and confirm what changed. For scoring and denominator detail across your program, use the customer satisfaction metrics guide.
