CSAT Survey: 8 Questions, the Formula and When to Send
A CSAT survey is the one question you ask right after you’ve helped someone. It only works if it lands while the conversation is still fresh, and if someone reads the bad answers.
Get both right and it takes a customer five seconds to answer and shows you which calls, texts and emails are going wrong while you can still fix them.
The quick answer: A CSAT (customer satisfaction) survey asks a customer to rate one support interaction, usually from 1 to 5. Your CSAT score is the share who answered 4 or 5: satisfied responses ÷ total responses × 100. Send it once, as soon as the conversation is resolved, in the channel the customer used: a text a few minutes after a call, an email when the thread is resolved. Carly can run the whole loop for you: a workflow texts or emails the survey when a support call, text or email wraps up, logs every score to a data table and flags low scores to a person on your team.
CSAT survey questions to copy
Eight templates, from the classic CSAT question to the two metrics people confuse it with. Swap in your own business name.
| Question type | Copy-paste wording | Scale | Use it for |
|---|---|---|---|
| CSAT (classic) | “How satisfied were you with the help you got today?“ | 1 (very unsatisfied) to 5 (very satisfied) | Your headline CSAT score |
| CSAT by text | ”How did we do today? Reply with a number from 1 (poor) to 5 (great).“ | 1 to 5, sent back as a reply | Surveys after a phone call or a text conversation |
| Thumbs up or down | ”Did we take care of you today?” with a thumbs-up and a thumbs-down button | 2 points | Chat widgets and email footers, where speed beats detail |
| Three faces | ”How was your experience?” with an unhappy, a neutral and a happy face | 3 points | In-app prompts and help-center pages |
| Resolution check | ”Did we solve your problem today?” | Yes or no | Tracking first-contact resolution next to CSAT |
| Open follow-up | After a 1 to 3: “What’s one thing we could have done better?” After a 4 or 5: “What did we get right?” | Free text | The reason behind the number |
| CES (effort) | “How easy was it to get your issue sorted?“ | 1 (very difficult) to 7 (very easy) | Finding friction: transfers, repeat contacts, long holds |
| NPS (for contrast) | “How likely are you to recommend us to a friend or colleague?“ | 0 to 10 | Loyalty to the whole company, asked quarterly rather than after every ticket |
Lead with one rating question. Ask the open follow-up only after the customer has answered it, and make it optional.
CSAT, CES and NPS measure different things. CSAT rates a single interaction. Customer Effort Score asks how hard the customer had to work to get an issue resolved, an idea popularized by the 2010 Harvard Business Review article “Stop Trying to Delight Your Customers.” NPS, the loyalty metric behind Bain & Company’s Net Promoter System, takes the percentage of promoters (9 or 10) and subtracts the percentage of detractors (0 to 6). A customer can give you a 5 for a friendly call and still score you a 6 on NPS because your prices went up. Run CSAT after support conversations and keep NPS for a quarterly pulse.
How to calculate your CSAT score
CSAT = (number of 4 and 5 ratings ÷ total ratings) × 100
Only count people who answered. Surveys nobody answered belong in your response rate, not your score:
Response rate = (surveys answered ÷ surveys sent) × 100
A worked example: Claire’s support team
Claire runs a five-person support team at a 40-person HVAC company in Denver. In September her team resolved 1,240 calls, texts and emails and surveyed every one. 310 customers answered, a 25% response rate.
| Rating | Responses |
|---|---|
| 5 (very satisfied) | 212 |
| 4 (satisfied) | 51 |
| 3 (neutral) | 18 |
| 2 (unsatisfied) | 11 |
| 1 (very unsatisfied) | 18 |
| Total | 310 |
Satisfied responses: 212 + 51 = 263. CSAT: 263 ÷ 310 × 100 = 84.8%.
Some tools report the average rating instead: (212×5 + 51×4 + 18×3 + 11×2 + 18×1) ÷ 310 = 4.38 out of 5. Pick one method and stick with it, because a “4.4” and an “85%” from the same month look like two different results.
The headline number hides the story. Here is the same month split by channel:
| Channel | Responses | Satisfied (4 or 5) | CSAT |
|---|---|---|---|
| Phone calls | 120 | 92 | 76.7% |
| Text messages | 90 | 80 | 88.9% |
| 100 | 91 | 91.0% |
Claire reads the comments on the phone 1s and 2s and finds the same complaint over and over: calls after 6 pm went to voicemail, and nobody called back until the next afternoon. Coaching won’t fix that. Answering the 7 pm call will, which is what Carly for teams does when nobody on the team can pick up. Our after-hours answering service guide covers the options.
When to send a CSAT survey
Send it once, when the conversation is resolved, not after every message. A survey that shows up the next day is competing with everything else in the customer’s inbox, about a conversation they’ve half forgotten.
- After a phone call: text the caller a few minutes after the call ends, from the number they called. They recognize it and can answer from wherever they are.
- After a text conversation: ask in the same thread once the issue is sorted. Don’t switch them to email just for the survey.
- After an email thread: send it when you mark the thread resolved, or a few hours later if customers often come back with “actually, it’s still broken.” For comparison, Zendesk’s default automation sends its survey 24 hours after a ticket is solved.
- Mind the clock: if a call ends at 10:40 pm, hold the text until morning. The FCC’s telemarketing rule draws the line at 8 a.m. to 9 p.m. in the recipient’s local time. A survey isn’t a sales message, but it’s a sensible window for any business text.
- Know when to skip it: spam, wrong numbers, abusive contacts and conversations that are still open. Cap surveys per customer too (once a week is a reasonable limit), so someone with three questions on Monday doesn’t get three surveys.
CSAT surveys channel by channel
Put the rating in the email body, not behind a “take our survey” button. Five links, one per score, so a single click records the answer and opens the optional comment box. Send it from the address the customer has been writing to, in the same thread where your tool allows it, with a subject line that names their issue.
Subject: How did we do with your furnace repair?
Hi Josh,
Thanks for getting in touch about your furnace. How satisfied
were you with the help you got?
[1] [2] [3] [4] [5]
1 = very unsatisfied, 5 = very satisfied
One click is all it takes. If anything still isn't right,
reply to this email and it comes straight to me.
ClaireText message after a call
The phone channel is where most survey programs fall apart. A keypad survey at the end of the call loses everyone who hangs up first. A text a few minutes later reaches them anyway, and replying with one digit takes a second.
Hi Emily, it's Claire at [Company]. Thanks for calling about your AC today. How did we do? Reply 1 (poor) to 5 (great). Reply STOP to opt out.Keep it under 160 plain characters (no emoji) so it arrives as a single message, name your business in the first line, and honor STOP. When a customer replies “2, still no heat,” that reply needs a person today, not a row in next month’s report. For number registration and opt-out basics, see our business texting guide.
Chat and in-app
Show the thumbs or three faces at the end of the chat, before the window closes. Once it closes, you’ve lost the moment.
How to raise your CSAT response rate
- One question first. The rating and nothing else. Anything more is optional and comes after.
- Same channel, same sender. Text after calls and texts, email after emails, from the number or address the customer already knows.
- Send it fast. Minutes after a call, not in tomorrow’s batch.
- Make it personal. “How did Claire do with your furnace?” reads like part of the conversation. “Please rate your recent interaction” reads like a form letter.
- Make it one tap. A reply digit or a one-click link. No logins, no landing page with ten fields.
- Show that someone reads it. Tell customers what changed because of their feedback: “You told us evening calls went to voicemail, so now they get answered.”
- Don’t bribe or coach. Gift-card draws and “a 5 would really help me out” inflate the score until it stops telling you anything.
What to do with a low CSAT score
A 1 or 2 is a customer telling you something went wrong, often one they haven’t told you anywhere else. Treat it as a new support conversation, not a data point.
- Route it to a person the same day. The team lead, or whoever owns the account, with the conversation and the customer’s comment attached.
- Read the whole conversation before replying. Find out what actually went wrong: wait time, a wrong answer, a transfer, a policy or the product itself.
- Reply personally, and reopen the issue if it isn’t fixed. A phone call often lands better than another email.
- Tag the reason. After a month, the tags tell you whether to fix staffing, training, a policy or the product.
- Don’t ask for a better score. Fix the problem and let the next survey show it.
Treat 3s as unsatisfied as well. They don’t count toward CSAT, and a “neutral” after a support conversation usually means the problem only half went away.
How to automate CSAT surveys with Carly
Sending surveys, collecting replies and chasing the low scores is exactly the work that slips when the team is busy. Carly runs it as a workflow you describe in plain English. Carly for teams answers your business line and texts when the team can’t, and logs every call and text in one team inbox alongside email, so surveys go out after the conversations Carly handled and the ones your team took over.
- Get started with Carly and connect your support channels. Connect your support inbox in Gmail or Outlook, and set up your business number with the Carly team (the one you already have, or a new one).
- Write the survey workflow. For example: “When a support call ends and the summary says the issue was resolved, text the caller from our business number: [your template].” For email: “Every day at 4 pm, find the support emails we resolved today and email each customer this survey: [your template].” Carly workflows trigger on call ended, call summarized, text received, new email and schedules, and a call workflow sees the outcome, summary and transcript.
- Log every score to a data table. “When a customer replies to a survey with a number, add a row to the CSAT table: date, customer, channel, who handled it, score and comment.” Carly creates the data table itself, no spreadsheet needed, and can update a CSAT column on the customer’s record in its built-in CRM.
- Flag low scores to a person. “If the score is 1 or 2, email Josh the conversation summary and the customer’s comment.” Josh can take the thread over from the team inbox, and Carly steps back.
- Get the number every week. “Every Monday at 9 am, email me last week’s CSAT by channel, the response rate and every score under 3.”
Carly connects to thousands of apps, so scores can also land in the helpdesk, CRM or Slack channel your team already uses. Pricing: free Zapier-style workflows; AI agents from $35/month. For the wider field of support tools, see our roundup of the best AI agents for customer support.
What’s a good CSAT score?
There’s no single pass mark that fits every business, and a number borrowed from another industry tells you little about yours.
The broadest public benchmark is the American Customer Satisfaction Index, which measures US satisfaction with more than 400 companies in over 40 industries from about 200,000 interviews a year. Its first-quarter 2026 release reported US satisfaction falling with customer complaints at record levels, and its second-quarter release on August 11 reported a sharp decline. The ACSI score comes out of a statistical model, so it isn’t the same thing as the percentage of 4s and 5s on your survey. Read it as the mood your customers are in, not as a target. Our customer service statistics roundup has more on what customers expect.
The benchmark that matters is your own: last month’s CSAT, by channel and by who handled the conversation. If phone sits 14 points below email, as it did for Claire, that gap is your to-do list.
FAQ
What is a CSAT survey?
A short survey sent after a customer interaction that asks how satisfied the customer was, usually on a 1 to 5 scale. CSAT stands for customer satisfaction. Most are one rating question plus an optional comment.
How do you calculate CSAT?
Divide the number of 4 and 5 ratings by the total number of ratings and multiply by 100. For example, 263 satisfied out of 310 responses is 84.8%. Unanswered surveys don’t count toward the score; they go into your response rate.
What’s the difference between CSAT, NPS and CES?
CSAT rates one interaction. NPS asks how likely someone is to recommend you on a 0 to 10 scale and measures loyalty to the whole company. CES asks how easy it was to get an issue resolved. Use CSAT after support conversations, CES to find friction and NPS quarterly.
When should you send a CSAT survey?
As soon as the conversation is resolved, in the channel the customer used: a text minutes after a call, an email when the thread is resolved. Hold late-night texts until morning, and don’t survey the same customer more than about once a week.
How many questions should a CSAT survey have?
One required rating question and one optional open question. The rating gives you the score and the comment gives you the reason. Anything beyond that belongs in a separate, occasional survey.
Can you send a CSAT survey by text message?
Yes, and after a phone call it’s usually the best channel. Send it from the number the customer called, name your business, ask for a reply from 1 to 5 and include “Reply STOP to opt out.” Carly sends it automatically when a support call or text is resolved, logs each reply and flags the 1s and 2s to your team.
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
