CSAT, the customer satisfaction score, is the percentage of customers who rated a conversation, a purchase or a product as satisfied on a short survey. It is calculated by dividing the satisfied responses by the total responses and multiplying by 100, so 84 satisfied answers out of 120 give a CSAT score of 70 percent. Support teams ask for the rating when a conversation closes and read the score per teammate and per inbox.
CSAT is satisfied responses divided by total responses, multiplied by 100. A response counts as satisfied when it falls in the top two points of a five-point scale, or the top one of a three-point scale. Ask once, right after the conversation closes, with one question and one optional comment, and expect a minority of customers to answer. Read the score beside the response count and the comments, and treat it as a measure of the last conversation rather than of the relationship.
What CSAT means
CSAT stands for customer satisfaction, and the CSAT score is the share of survey responses that were positive. The survey asks one question about a specific moment, usually a conversation that just closed, in the form of a rating from very dissatisfied to very satisfied. Because the question is about one moment, the score moves with the quality of that conversation and the mood of the person answering it. That makes it a transactional measure, in contrast to relationship measures such as the net promoter score, which ask about the company as a whole.
The score has become standard in support teams because it is simple to collect and simple to explain. A manager can say that 86 percent of customers who rated a reply last month were satisfied with it, and a director can understand the sentence without a definition. The simplicity is also the limit, since the score says nothing about the customers who did not answer and nothing about why the satisfied ones were satisfied.
How to calculate a CSAT score
The formula is the number of satisfied responses divided by the total number of responses, multiplied by 100. On a five-point scale the satisfied responses are the fours and fives, and on a three-point scale of bad, okay and good they are the goods. Everything below the line counts in the total but not in the satisfied count, which is why the score falls when customers choose the middle option.
A worked example makes the arithmetic plain and gives the response rate as well. A team closed 400 conversations in a month and sent a survey after each one. It received 120 responses, of which 64 were fives, 20 were fours, 16 were threes, 12 were twos and 8 were ones. The satisfied responses are 64 plus 20, which is 84, and 84 divided by 120 is 0.70, so the CSAT score is 70 percent. The response rate, 120 out of 400, is 30 percent, and it is reported next to the score.
| Rating | Responses | Counted as satisfied |
|---|---|---|
| 5, very satisfied | 64 | Yes |
| 4, satisfied | 20 | Yes |
| 3, neutral | 16 | No |
| 2, dissatisfied | 12 | No |
| 1, very dissatisfied | 8 | No |
| Total | 120 | 84 satisfied, so CSAT is 70 percent |
Some teams report the average rating instead, which for the example is 4.0 out of 5. The average is useful for spotting a shift in the middle of the distribution, but it hides the share of unhappy customers. Two fives and two ones average the same as four threes. The percentage is the standard CSAT score, and the average is a companion figure rather than a replacement.
CSAT scales: emoji, numbers and stars
| Scale | Options | Counted as satisfied | Where it fits |
|---|---|---|---|
| Three faces | Sad, neutral, happy | Happy | Email and WhatsApp, where one tap is all a customer will give |
| Five points | 1 to 5, or very dissatisfied to very satisfied | 4 and 5 | Email surveys and help desks, and the most common form |
| Five stars | 1 to 5 stars | 4 and 5 stars | Apps and website chat, where the star is familiar |
| Thumbs up or down | Good or bad | Good | Website chat and help center articles, where a single tap suits the moment |
| Seven or ten points | 1 to 7 or 1 to 10 | The top two or three, stated in advance | Research surveys rather than conversation ratings |
The scale changes the score, so a team picks one and keeps it. A three-point scale produces higher scores than a five-point one, because the single neutral option absorbs fewer answers than the three and the four do on five points. A comparison between two teams, or two months, is only fair on the same scale with the same definition of satisfied.
When to ask for a CSAT rating
The rating is asked once, when the conversation closes, while the exchange is fresh and the customer knows what they are rating. Asking after every reply trains customers to ignore the request and punishes long conversations with several low ratings for one problem. Asking days later gets a rating of the memory of the conversation rather than the conversation itself. A sensible cap is one request per customer per week, so a customer who had three conversations is asked about one of them.
The request itself is short: one question, the scale, and an optional comment box. On email it sits under the closing reply or arrives as a separate message within an hour of the close. On WhatsApp or website chat it appears as the next message. Customer satisfaction survey email has templates for support, account management, complaints and onboarding.
What response rate to expect
Only a minority of customers answer a satisfaction survey, and the share varies with the channel, the length of the request and how soon after the close it arrives. Response rates are not comparable between companies, so the useful figure is the team's own rate over time, read beside the score. A score that rose while the response rate fell is suspect, since the customers who stopped answering may be the unhappy ones. A sudden change in either number after a change to the survey points to the survey rather than the service.
Two practices raise the rate without biasing the score. The request is sent as part of the closing reply rather than as a separate message. The rating is a single tap that opens a comment box only after the choice. Reminders are left out, because a second request for a rating irritates the customers who chose not to answer the first.
CSAT benchmarks
Published CSAT benchmarks come from help desk vendors, who average the scores of their own customers, and from survey firms. Those averages depend on the scale, the industries and the vendor's customer base, so a team's score cannot be read against them with confidence. This guide quotes no benchmark number for that reason. The American Customer Satisfaction Index, a national index on a 0 to 100 scale, measures satisfaction with companies and industries by interview. Its interview method differs from a one-click conversation rating.
The useful comparison is internal rather than against a published average. A team compares this month with last month, one inbox with another, and one teammate with the team's figure, all on the same scale. A score that is stable over three months, with a response rate that is also stable, is the baseline, and the question each month is what moved it.
How to read CSAT per teammate and per inbox
Per teammate, the score shows how each person's replies land, and it is read with the number of responses behind it. A teammate with 90 percent on 10 responses and one with 82 percent on 120 responses are not ranked by the two percentages. The first figure moves by ten points with one rating. A fair rule is to compare teammates only above a floor of responses, such as 30 in the period, and to read the comments for anyone below it.
Per inbox, the score separates channels and kinds of work. A billing inbox often rates lower than a sales inbox because the subject is a problem, and the fair comparison is the billing inbox against its own past months. An inbox tool that credits each rating to the teammate who sent the last reply and to the inbox the conversation lived in produces both views from the same data. That is how the figure is reported in a shared inbox's analytics.
What CSAT does not tell you
The score measures the last conversation and the customers who chose to rate it. It says nothing about customers who stopped writing, which is the earlier warning for an account team, and it says little about whether the customer will stay. A customer can be satisfied with every reply and still leave because the product did not do what they needed. The comments carry more of the explanation than the score, and a team that reads only the number misses them.
For an account team, three measures sit beside it. They are whether the promises made in email were kept on time, how many client questions are open and how old the oldest is, and which accounts have gone quiet. Customer success metrics beyond response time sets those out, and CSAT vs NPS covers the relationship measure most often paired with CSAT.
Where Sektra fits
Sektra sends a satisfaction survey when a conversation in a shared inbox closes, on email, WhatsApp and website chat, with a cap of one request per customer in a week. The ratings and comments show in the analytics per teammate and per inbox, beside first response time, resolution time and service level results. A low rating can be read with the conversation it came from. Lens, the relationship panel beside each thread, keeps the customer's history and open commitments, which supplies the context a score leaves out. Help desks such as Zendesk and Freshdesk offer a survey of the same kind on their middle or upper plans. The choice between tools rests on the channels and the reporting.
What to do with the score
Set the scale and the definition of satisfied in writing, send the request once at the close, and report the score with the response count every month. Read every rating of one or two within a day and reply to it, since the reply changes the customer's view more than the survey did. Compare teammates only above a response floor, compare inboxes against their own history, and read the comments before the number. Treated this way, CSAT is a reliable measure of how replies land, and it leaves the measurement of the relationship to the metrics built for it.
Common questions
What does CSAT stand for?
CSAT stands for customer satisfaction. The CSAT score is the percentage of survey responses that rated a conversation, purchase or product as satisfied, usually the top two points on a five-point scale.
What is a good CSAT score?
There is no reliable universal number, because published averages depend on the scale, the industry and the vendor's customer base. A team sets its baseline from three months of its own scores on one scale, then judges each month against that baseline and the response rate behind it.
How is CSAT calculated?
Divide the number of satisfied responses by the total number of responses and multiply by 100. On a five-point scale the satisfied responses are the fours and fives. For example, 84 satisfied responses out of 120 give a CSAT score of 70 percent.
What is the difference between CSAT and NPS?
CSAT measures satisfaction with one moment, such as a support conversation, right after it. NPS measures how likely a customer is to recommend the company, on a 0 to 10 scale, and is asked a few times a year. CSAT belongs to the support team and NPS to whoever owns the relationship.
