CSAT: Measuring Customer Satisfaction Score

·

customer satisfaction score CSAT

Not sure what your survey results really say? Your customer satisfaction score CSAT is the percentage of customers who rate an experience positively, and the table below shows what counts as good.

CSAT is one of the core customer experience metrics, and this page covers how to calculate it, read it and act on it.

Item Value
Typical good CSAT range 70% to 85%
Ratings that count as satisfied 4 or 5 out of 5
Worked example 75 satisfied responses out of 100 respondents = 75%
Full-service restaurants (ACSI benchmark, 2024) 84
Social media (ACSI benchmark, 2024) 74

The figures come from IBM’s CSAT guide and SurveyMonkey’s CSAT guide. Ranges and benchmarks can change, so check the sources before you rely on them.

What the customer satisfaction score measures

A customer satisfaction score indicates how satisfied a customer is with a company’s product or service, as IBM’s guide describes it. It accounts for a specific interaction a person has with the business, which makes it a close-up reading rather than a verdict on the whole relationship.

The survey usually asks, “How satisfied were you with your experience?” Respondents answer on a scale of 1-3, 1-5 or 1-10, or by percentage, with 100% being fantastic and 0% being very bad, according to the same IBM guide.

How to calculate CSAT

The arithmetic is short. IBM describes it in four steps:

  1. Collect survey responses on your chosen scale.
  2. Add up the positive responses, meaning ratings of 4 or 5 out of 5.
  3. Divide that sum by the total number of responses collected.
  4. Multiply the result by 100 to get a percentage.

SurveyMonkey gives a clean example: if you received 75 satisfied responses out of 100 respondents, 75% of customers were satisfied. Which ratings count as satisfied depends on the scale, and the same guide lists them like this:

  • On a scale of 1 to 5: 4 and 5
  • On a scale of 1 to 7: 5, 6 and 7
  • On a scale of 1 to 10: 6, 7, 8, 9 and 10

HubSpot shows a smaller case: with 25 total responses and 15 positive ones, the score is 60%. A result like that suggests most people are satisfied, yet it leaves clear room to improve.

A person wears a large yellow smiling face with black glasses against a red background.

What is a good CSAT score?

SurveyMonkey says a good score typically ranges from 70% to 85%, though the ideal figure varies by industry, as its CSAT guide explains. HubSpot places the typical range at 75% to 85%, and notes that 75% means three out of every four customers gave a positive score.

SurveyMonkey also breaks scores into bands:

  • 0-50%: needs improvement
  • 50-70%: fair
  • 70-85%: good
  • 85-100%: excellent

Industry benchmarks

A flat number hides sector differences, so SurveyMonkey advises benchmarking against your industry’s average. IBM cites American Customer Satisfaction Index (ACSI) figures that show the spread:

  • Full-service restaurants: 84 (2024)
  • Food manufacturing: 82 (2023)
  • Banks: 80 (2024)
  • Airlines: 77 (2024)
  • Social media: 74 (2024)

When to send a CSAT survey

The feedback request works best at defined moments, so the answers tie back to a specific interaction, per IBM. Typical moments include:

  • During onboarding and in the postpurchase stages of the customer cycle.
  • Before a subscription renewal; the sweet spot is six months out, which leaves time to act on the feedback.
  • At the close of a support call, as a single satisfaction question or a full survey depending on the interaction.

Because the survey is brief, you can repeat it across several touchpoints and spot where the experience slips.

Strengths and limits of CSAT

IBM notes that CSAT surveys are typically short and simple to complete, which keeps response friction low. The trade-offs are real, though, and the same guide lists three:

  • Outside influences, such as current events or a customer’s personal situation, can skew a score.
  • The survey is not in-depth enough to explain why people are satisfied or dissatisfied.
  • Customers who had an exceptionally good or bad interaction are more likely to respond, which slants results toward the extremes.

How CSAT differs from NPS and CES

IBM says CSAT covers a specific product or service and does not give a full, ongoing look at long-term sentiment. For the loyalty angle, this walkthrough on how to calculate NPS covers a score built on a likelihood-to-recommend question.

For ease of use, the customer effort score measures how easy it is to complete a task. Used together, the three metrics give an organization several reference points, according to IBM.

Ways to improve a customer satisfaction score

Improvement starts with feedback you already have. IBM suggests four habits:

  1. Listen actively: pay attention to what customers say and act on it.
  2. Stay proactive: reaching out before there is an issue beats only interacting when there is a problem.
  3. Weigh outside influences when you read results or before you run the survey.
  4. Keep up with new technology and keep improving the digital experience.

SurveyMonkey adds that addressing feedback, improving service quality and streamlining processes are the practical levers, per its guide.

Frequently asked questions

Can a business reach a perfect CSAT?

Show answer

Both SurveyMonkey and HubSpot call a perfect or near-perfect result difficult. Aim to keep improving rather than chase a flawless figure.

How long should a CSAT survey be?

Show answer

It depends on the interaction. After a support call, IBM says an organization might ask just one satisfaction question or run the full survey.

Does CSAT replace NPS?

Show answer

No. The metrics answer different questions, so many organizations track all three and compare themselves with competitors using more than one measure, as IBM points out.

Which rating scale should I use?

Show answer

The metric depends on the organization and how it decides to measure overall customer experience, according to IBM. Whatever you pick, decide in advance which ratings count as satisfied and keep that rule fixed.