CSAT: How to Calculate It, 20 Survey Question Examples, and a Ready-Made Template
Add up the survey scores, divide by the total, and you get 4.1 out of 5, which reads as 82% satisfaction. The CSAT of that same survey is 71%. The gap has a simple explanation. CSAT counts only the people who answered 4 or 5, and that is the standard definition of CSAT. This page covers the math and the rest of the job: a calculator, 20 ready-to-use CSAT survey questions organized by touchpoint, 3 survey templates you can copy, and an industry benchmark table with sources.
- CSAT calculator
- How to calculate CSAT
- 20 CSAT survey question examples
- 3 ready-made survey templates
- What is a good CSAT score? Benchmarks by industry
“I want a satisfaction survey” is a request that reaches our support team regularly: a car wash that wants to hear from the customer after handing back the car, someone building one NPS and two CSAT surveys and wanting every response to become a task for the team, a company that wants to hear from its own employees. The math and the questions are the same in all three cases.
CSAT calculator
Type in how many responses each score received. The calculator applies the standard formula (scores of 4 and 5 divided by the total) and shows, next to it, the average that gets mistaken for CSAT.
1 to 5 scale. Responses per score:
71 of 100 responses scored 4 or 5. The 4.1 average (82% of the maximum score) is not the CSAT.
If your scale has 3, 7 or 10 points, the table in the next section shows what counts as a satisfied response on each one. The division is the same.
How to calculate CSAT
The formula is one line:
CSAT (%) = satisfied responses ÷ total responses × 100
On a 1 to 5 scale, “satisfied” means the responses of 4 and 5. Survey researchers call this top-2-box, the two boxes at the top of the scale. Scores of 1, 2 and 3 only enter the denominator.
A worked example, with 250 responses to “how would you rate the support you just received?”:
| Score | Responses | Counts toward CSAT? |
|---|---|---|
| 5 | 130 | Yes |
| 4 | 48 | Yes |
| 3 | 45 | No |
| 2 | 22 | No |
| 1 | 5 | No |
| Total | 250 | 178 |
CSAT = 178 ÷ 250 × 100 = 71.2%.
The average of those same 250 scores is 4.1, and this is the number that gets reported as CSAT by mistake. The average treats a 3 as “60% satisfied”, but a 3 is a neutral score. Two surveys can share an average and have opposite CSATs: half the responses at 5 and half at 3 gives an average of 4.0 and a CSAT of 50%; every response at 4 gives an average of 4.0 and a CSAT of 100%. Report the percentage, and show the score distribution next to it whenever the number surprises someone.
What counts as satisfied changes with the scale:
| Scale | Counts as satisfied | Note |
|---|---|---|
| 3 points (poor, fair, good; or faces) | Only the top point | This only holds when the three points cover dissatisfied, neutral and satisfied. A street-food stand that asked our support for help used “excellent, good and fair”, a scale with no negative point: counting only “excellent” there measures excellence, not satisfaction. |
| 5 points (1 to 5) | 4 and 5 | The market standard for CSAT |
| 7 points (1 to 7) | 6 and 7 | More sensitive to small shifts; the middle labels get hard to name |
| 10 points (1 to 10) | No single cutoff: 9 and 10 in the stricter reading, 8 to 10 in the more common one | Decide before the first collection and don’t change it. Moving the cutoff midway breaks the historical series |
Two rules hold for any scale: label both ends (1 = very dissatisfied, 5 = very satisfied) and keep the same scale in every round. Otherwise March’s number is not comparable with April’s.
What CSAT measures (and what it doesn’t)
CSAT stands for Customer Satisfaction Score. It measures satisfaction with a specific interaction, right after it happens: a support conversation, a purchase, a delivery, a class, an appointment. The question is always a variation of “how would you rate [the interaction]?” on a short scale, and the result is the percentage of satisfied responses.
CSAT doesn’t measure loyalty or intent to recommend; that is what NPS, the Net Promoter Score, is for. NPS asks, on a 0 to 10 scale, how likely someone is to recommend the company. It also doesn’t explain why the score dropped, which is why every CSAT survey needs an open-ended question next to the rating. Use CSAT when you want to track the quality of one touchpoint over time and react quickly to a low score.
20 CSAT survey question examples by touchpoint
These CSAT survey questions are grouped by touchpoint: pick yours, copy the rating question and one or two supporting questions. A good CSAT survey has 1 to 4 questions. The answer format is in parentheses; questions 1, 6, 11 and 15 are the CSAT questions proper, and the others explain the score. For a larger bank organized by survey goal, see the 100+ survey question examples.
After a support interaction
- How would you rate the support you just received? (1 to 5, very dissatisfied to very satisfied)
- After this conversation, do you still need help with the same issue? (yes / no)
- How would you rate the time you waited before being helped? (1 to 5)
- Did you have to repeat the same information to more than one person? (yes / no)
- What was missing for this to deserve a 5? (open; show only to scores of 1 to 3)
After a purchase or delivery
- How would you rate your shopping experience today? (1 to 5)
- Did your order arrive within the promised timeframe? (yes / no / hasn’t arrived yet)
- Does the product match what was advertised? (1 to 5, “not at all” to “completely”)
- Which step of this order took longer than you expected? (multiple choice: payment confirmation / shipping / delivery / none)
- Would you buy from us again? (yes / maybe / no)
Product and usage
- How would you rate [product] today? (1 to 5, very dissatisfied to very satisfied)
- Did the feature you used today do what you needed? (yes / partly / no)
- How would you rate the ease of use of [feature]? (1 to 5, very difficult to very easy)
- What should the product do that it doesn’t do yet? (open)
After a service, class or appointment
- How would you rate today’s [class / appointment / session]? (1 to 5)
- Was what we agreed on delivered? (yes / partly / no)
- Did you leave with what you needed for the next step? (yes / no; on no, open a text field)
- How easy was it to book and get here? (1 to 5, very difficult to very easy)
Closing, for any touchpoint
- What weighed most in the score you gave? (open, everyone)
- Can we follow up with you about this answer? (yes, with an email or phone field / no)
Question 19 is the most important one after the rating. With a few dozen open answers you can see the pattern by reading them one by one; with a few hundred, manual reading stops scaling, and that is where AI analysis comes in (more on that at the end of the page).
3 ready-made CSAT survey templates
Each template below is a whole survey: screens, questions, formats and logic. Copy it into your form builder, or paste the prompt into Yay! Forms to generate the survey in one step.
Template 1: after a support interaction (3 questions)
- Welcome screen: “It takes 20 seconds. How was today’s support?”
- Question 1: How would you rate the support you just received? (1 to 5 scale; 1 = very dissatisfied, 5 = very satisfied)
- Question 2, only for scores of 1 to 3: What was missing for this to deserve a 5? (free text)
- Question 3: Can we follow up with you about this answer? (yes / no; on yes, an email or phone field)
- Thank-you screen: “Thank you. Your answer goes straight to the team that helped you.”
Template 2: after a purchase or delivery (4 questions)
- Welcome screen: “One question about your order. It takes 20 seconds.”
- Question 1: How would you rate your shopping experience? (1 to 5 scale)
- Question 2: Did your order arrive within the promised timeframe? (yes / no / hasn’t arrived yet)
- Question 3, only for people who already received the order: Does the product match what was advertised? (1 to 5 scale)
- Question 4: What weighed most in the score you gave? (free text, optional)
- Thank-you screen: “Thank you. If something went wrong, we’ll fix it: reply to this email or message us.”
Template 3: after a class, appointment or service (4 questions)
- Welcome screen: “Four questions about today’s session. It takes under a minute.”
- Question 1: How would you rate today’s [class / appointment / session]? (1 to 5 scale)
- Question 2: Was what we agreed on delivered? (yes / partly / no)
- Question 3: Did you leave with what you needed for the next step? (yes / no; on no, open a text field “What was missing?”)
- Question 4: What could we do better next time? (free text)
- Thank-you screen: “Thank you. We read every answer before the next session.”
In all three templates, the survey goes out through the channel where the interaction happened: a link in the chat after support, the order email, a QR code at the counter or in the room. For the post-purchase one, trigger it on the promised delivery date rather than on the delivery confirmation: the customers still waiting are the ones with the most to say, and tying the send to “order delivered” erases them from the count.
What is a good CSAT score? Benchmarks by industry
It depends on the industry and the touchpoint, and the most useful number is your own CSAT from last month, on the same touchpoint and the same scale. Still, an external reference helps you calibrate. The most transparent source is the ACSI (American Customer Satisfaction Index), which publishes industry indexes based on interviews with U.S. consumers. The ACSI is a 0 to 100 index with its own methodology, different from the percentage of 4s and 5s. It is useful for comparing industries with each other and seeing the order of magnitude. Concluding that your 78% CSAT “beat the banks” is reading the index wrong.
| Industry | ACSI (0 to 100) | Study |
|---|---|---|
| Full-service restaurants | 82 | Restaurants 2026 |
| Financial advisors | 82 | Finance 2026 |
| Banks | 80 | Finance 2026 |
| Specialty retailers | 80 | Retail 2026 |
| Online retailers | 79 | Retail 2026 |
| Quick-service restaurants | 79 | Restaurants 2026 |
| Supermarkets | 78 | Retail 2026 |
| Lodging | 77 | Travel 2026 |
| Wireless phone service | 77 | Telecom 2026 |
| Video streaming | 77 | Entertainment 2026 |
| Hospitals | 77 (2025) | Health care |
| Airlines | 76 | Travel 2026 |
| Health insurance | 76 (2025) | Insurance 2025 |
| Food delivery | 75 | Restaurants 2026 |
| Internet service providers | 73 | Telecom 2026 |
| AI platforms | 73 | AI Platforms 2026 |
| Subscription TV | 72 | Entertainment 2026 |
| Federal government | 70.4 (2025) | Government 2025 |
| U.S. national average, Q2 2026 | 76.1 | Overall index |
The industries in the table land between 72 and 82, a 10-point gap from the worst to the best, with the federal government below all of them at 70.4. The same score means different things in food delivery (industry at 75) and in a full-service restaurant (industry at 82).
Several countries publish their own national satisfaction indexes, a number of them built on the ACSI methodology, and they measure the overall relationship with a brand rather than a single interaction.
How many responses you need
How much a 71% CSAT is worth depends on how many people answered. The margin of error, the range around the result where the true value most likely sits, puts a number on that. The table below uses the standard formula for proportions, at 95% confidence and with the most conservative assumption (a 50% proportion, which maximizes the margin):
| Responses | Margin of error (95%) | Reading of a 71% CSAT |
|---|---|---|
| 30 | ± 17.9 points | between 53% and 89% |
| 50 | ± 13.9 points | between 57% and 85% |
| 100 | ± 9.8 points | between 61% and 81% |
| 200 | ± 6.9 points | between 64% and 78% |
| 400 | ± 4.9 points | between 66% and 76% |
| 1,000 | ± 3.1 points | between 68% and 74% |
Two practical consequences. First: under 100 responses, a 5-point swing from one month to the next is inside the noise; don’t reorganize the team over it. Second: the margin assumes that the people who answered are a random sample of the people who were served. If only your happiest customers respond, no sample size fixes that, and the way to reduce the bias is to send the survey to everyone who was served, through the same channel and within the same window.
When to send it and how to raise the response rate
- Send it right away. The survey goes out on the same day as the interaction, through the channel where it happened. A support survey sent a week later measures memory, not support.
- One rating and one open question are enough. The longer a questionnaire’s announced length, the fewer people start and finish it, and the answers near the end get shorter and more uniform, as Galesic and Bosnjak’s web-survey experiment showed. A CSAT survey needs to fit in 20 seconds.
- Identify the respondent when you intend to react. Hidden fields with the order or ticket number tie the score to the case without asking the customer for anything. If the survey is anonymous by choice, say so on the welcome screen.
- Set a frequency cap. Surveying every interaction is the default, and it is what keeps the metric at the interaction level. If that means a frequent customer gets three surveys in a week, decide in advance whether to cap it (one per customer every 30 days, say) and record that the CSAT then covers a sample of interactions rather than all of them.
- Close the loop. A score of 1 answered by a person on the same day is worth more than the next survey round. Without someone on the other end, customers learn that the survey changes nothing and stop responding.
CSAT, NPS or CES?
CSAT, NPS and CES (Customer Effort Score) measure customer perception and answer different questions:
| Metric | Question | Scale | What it measures | Use it when |
|---|---|---|---|---|
| CSAT | How would you rate [the interaction]? | 1 to 5 | Satisfaction with a specific moment | You want to track a touchpoint and react quickly |
| NPS | How likely are you to recommend [the company] to a friend or colleague? | 0 to 10 | Loyalty and intent to recommend | You want one relationship metric, comparable with the market |
| CES | How easy was it to resolve your issue today? | 1 to 7 | Effort demanded from the customer | You want to reduce friction in support and self-service |
The three metrics complement each other. A common setup places CSAT after every support interaction and NPS once or twice a year, with the overall score followed by diagnostic statements on a Likert scale. Yay! Forms has a ready-made NPS field, with the 0 to 10 scale already labeled.
How to create a CSAT survey in Yay! Forms
The rating question uses the Opinion Scale field, set from 1 to 5 with labels at both ends, or the star rating field when the interaction calls for a more visual format. The open question uses the long text field, and conditional logic shows “what was missing?” only to people who scored 1 to 3.
None of this has to be built by hand. Paste one of the prompts on this page and Yay! Forms AI generates the survey with the questions, the options and the logic in place. You adjust the wording and publish it as a link, a QR code or an embed on your site.
After collection, the Summary shows how many responses each score received; the CSAT is the sum of 4s and 5s divided by the total, the same math as the calculator above. For the historical series, export the responses to a spreadsheet. The open answers go through AI analysis, which summarizes the most cited reasons and classifies sentiment without anyone reading them one by one.
To close the loop, integrations accept conditional expressions: the Slack, WhatsApp or email notification fires only when the score is 1 or 2, and the team gets back to the customer the same day. Notifications are available on the Plus and Professional plans; the 7-day free trial doesn’t ask for a credit card.
Frequently asked questions
What is CSAT?
CSAT (Customer Satisfaction Score) is a measure of customer satisfaction with a specific interaction, such as a support conversation, a purchase or an appointment. It is measured with a question like “how would you rate [the interaction]?” on a short scale, usually 1 to 5, and expressed as the percentage of satisfied responses (scores of 4 and 5).
How do you calculate CSAT?
Divide the number of satisfied responses by the total number of responses and multiply by 100. On a 1 to 5 scale, satisfied responses are the scores of 4 and 5. With 178 responses of 4 or 5 out of 250, the CSAT is 178 ÷ 250 × 100 = 71.2%. The calculator at the top of this page does the math for any score distribution.
What is the difference between CSAT and NPS?
CSAT measures satisfaction with a specific moment, right after it happens, on a 1 to 5 scale. NPS measures how likely someone is to recommend the company, on a 0 to 10 scale, and summarizes the relationship as a whole. CSAT is sent after each interaction; NPS, at longer intervals. Neither replaces the other.
What is a good CSAT score?
It depends on the industry and the touchpoint. Across the industries in the table above (ACSI studies from 2025 and 2026), indexes range from 72 to 82 on a 0 to 100 scale, with banks at 80 and full-service restaurants at 82. Since the ACSI is not the percentage of 4s and 5s, the most useful comparison is with your own history: the same touchpoint, month after month, on the same scale.
Is CSAT a percentage or an average score?
A percentage: the share of satisfied responses over the total. The average score (4.1 out of 5, for example) is a different metric, and the gap between the two changes direction with the distribution. Because the average gives partial credit to the 1s, 2s and 3s, it sits above the CSAT when low scores are common (4.1 reads as 82% of the top score next to a CSAT of 71%) and below it when most people stop at 4, which counts as 100% in the CSAT and 80% in the average. When comparing against benchmarks or other teams, use the percentage.