CSAT vs NPS
Differences, Formulas and When to Use Each
Customer Satisfaction Score (CSAT) and Net Promoter Score (NPS) are two of the most widely used ways of putting a number on customer experience. They answer different questions. CSAT tells you how a customer felt about one interaction. NPS tells you how they feel about your brand overall.
Many teams track both, so the useful question is which metric should inform which decision. This guide covers how to calculate each one, 2026 benchmarks, where Customer Effort Score fits, and how ecommerce support teams should use them.
What is the difference between CSAT and NPS?
CSAT measures how satisfied a customer is with a specific interaction, usually on a 1–5 scale and asked straight after it happens. NPS measures how likely a customer is to recommend your brand, on a 0–10 scale, as a read on the whole relationship. CSAT is a transactional metric and NPS is a relational one.
Here's how they compare side by side:
The question: CSAT asks "How satisfied were you with…?". NPS asks "How likely are you to recommend us to a friend or colleague?"
The scale: CSAT usually runs from 1 to 5. NPS runs from 0 to 10.
Timing: CSAT is sent straight after a touchpoint such as a support conversation or a delivery. NPS is sent periodically, or at a set point in the customer lifecycle.
What it measures: CSAT measures satisfaction with a moment. NPS measures loyalty to the brand.
The formula: CSAT is either the percentage of satisfied responses or the average score. NPS is the percentage of promoters minus the percentage of detractors.
Score range: CSAT runs from 0% to 100% (or 1 to 5 as an average). NPS runs from −100 to +100.
Best used for: CSAT is best for fixing specific interactions, agents and processes. NPS is best for tracking brand health and loyalty over time.
How do you calculate CSAT?
There are two common ways to calculate CSAT. The percentage method divides the number of satisfied responses (usually 4s and 5s on a 1–5 scale) by the total number of responses and multiplies by 100. The average method takes the mean of every score. A typical question is "How satisfied were you with the help you received today?"
Method 1: percentage of satisfied customers. This is the version Qualtrics sets out, and it gives a score between 0% and 100%:
CSAT = (number of satisfied responses (4s and 5s) ÷ total responses) × 100
Method 2: average score. Add up every score and divide by the number of responses. On a 1–5 scale this gives a score between 1 and 5, which some helpdesks show by default:
CSAT = sum of all scores ÷ total responses
Here's the same set of 400 survey responses calculated both ways:
220 customers scored 5, 120 scored 4, 30 scored 3, 20 scored 2 and 10 scored 1.
Percentage method: 340 of the 400 responses are a 4 or 5, so CSAT is 85%.
Average method: the scores add up to 1,720, and 1,720 ÷ 400 gives a CSAT of 4.3 out of 5.
The percentage method is easier to read at a glance and less affected by a handful of very low scores. The average method uses every response, so it picks up movement in the middle of the scale (a shift from 2s to 3s, for example) that the percentage method ignores. Either works if you stick with it.
There are other conventions too. The American Customer Satisfaction Index reports satisfaction on a 100-point scale, calculated as a weighted average of three survey questions. Before you compare your CSAT with anyone else's, check which method and scale they're using: 85% and 4.3 out of 5 can describe exactly the same customers.
How do you calculate NPS?
NPS asks one question: "How likely are you to recommend us to a friend or colleague?" Customers answer on a 0–10 scale. Those who score 9 or 10 are promoters, 7 or 8 are passives, and 0 to 6 are detractors. NPS is the percentage of promoters minus the percentage of detractors.
That's the method set out by the Net Promoter System at Bain & Company:
NPS = % promoters − % detractors
For example, take 1,000 responses: 520 promoters, 300 passives and 180 detractors. That gives 52% − 18% = an NPS of 34. Passives count towards the total but don't affect the score directly. Because detractors can outnumber promoters, NPS runs from −100 to +100.
NPS was introduced by Fred Reichheld of Bain in a 2003 Harvard Business Review article. It argued that "the best predictor of top-line growth can usually be captured in a single survey question". That claim made NPS a boardroom favourite, and it has been debated ever since.
What are the pros and cons of CSAT and NPS?
CSAT is quick and specific, and you can act on it at agent or process level. However, it reflects the outcome of an interaction as much as the quality of the service. NPS captures brand loyalty in a single question, but one number can't tell you what caused it.
CSAT pros:
It's tied to a specific interaction, so you can trace a low score back to an agent, a channel or a process.
CSAT cons:
It measures how the customer feels about the outcome, which isn't always how well your team performed. If a customer asks for a full refund on an item just outside its warranty, the agent who explains the policy correctly can still get a low score.
A customer may be happy with the result but frustrated by how much effort it took to get there, and a single score can't tell those apart.
NPS pros:
It reflects the whole experience: product, price, delivery, service and brand.
NPS cons:
The score alone doesn't explain itself. A customer who gives you a 5 might be unhappy with product quality, price, delivery time or a support conversation.
It moves slowly, so it's a poor tool for judging a single change.
When should you use CSAT instead of NPS?
Use CSAT straight after a specific touchpoint, such as a support conversation, a delivery or a return, when you want to improve that touchpoint. Use NPS periodically, for example every quarter or a set time after purchase, when you want to track loyalty and brand health over time.
In an ecommerce journey, that usually looks like this:
After a support ticket is resolved: CSAT.
After a delivery, return or exchange: CSAT, or Customer Effort Score if you want to measure friction.
Every three to six months, or after a customer's second order: NPS.
The right metric also depends on who's asking. Ines Castro from On made this point when she spoke to us about the metrics customer service leaders should track. A customer service team leader might focus on their team's CSAT or backlog, while a CEO might look at satisfaction because retention drives growth. At On, the one KPI every department knows is NPS.
Which metric predicts customer loyalty better, NPS or CSAT?
Neither metric has proved to be the single best predictor of loyalty. Bain's research links NPS to growth. Peer-reviewed studies, however, have found NPS no better than other satisfaction measures, and in one study simpler "top-box" versions of the recommend question did better. The most reliable approach is to track several measures against actual repeat purchase.
The case for NPS rests on Reichheld's research and Bain's follow-up work. The case against it comes mainly from academics. A study in the Journal of Marketing by Keiningham and colleagues used data from 21 firms and more than 15,500 interviews. It "fails to replicate" the "clear superiority" of NPS over other measures (Keiningham et al., 2007). A more recent study of 193,220 responses across seven sportswear brands found that "'top-box' metrics perform better" (Baehre et al., 2022). Top-box metrics are the share of customers giving the highest scores (in that study, on the likelihood-to-recommend question), the same logic as the percentage version of CSAT.
Customer service leaders have been questioning NPS too. In 2021, Gartner predicted that more than 75% of organisations would abandon NPS as a measure of success for customer service and support by 2025. At the time, 58% of service leaders were still required by their executives to measure it (Gartner, 2021).
The practical answer is to test both metrics against your own data. Match survey scores to customer records and see which one moves with repeat purchase rate. Club L London has seen this link in practice. The team tracks CSAT alongside its Trustpilot rating, and sees a direct correlation between improved customer sentiment and retention, with customers coming back more often. It's a good example of customer service driving retention.
What is a good CSAT or NPS score?
For US online retailers, the American Customer Satisfaction Index puts customer satisfaction at 79 out of 100 in 2026, with Amazon, Chewy and Nordstrom tied at the top on 82. US consumer NPS averages range from about 16 to 34 depending on the industry. Your own trend, and how you compare with direct competitors, is the more useful benchmark.
The ACSI Retail and Consumer Shipping Study 2026 (31,293 surveys through 2025) breaks satisfaction down by stage:
Online retailers overall: 79
Quality of delivery: 83
Ease of making returns: 80
Across the US economy, the national ACSI score was 76.7 in the first quarter of 2026 (ACSI, 2026).
NPS varies more by sector. Qualtrics XM Institute's 2024 consumer ratings put grocers highest with an average NPS of 34.3 and car rental lowest at 15.8, across 22 industries (Qualtrics XM Institute, 2025). Any positive NPS means more promoters than detractors. Beyond that, compare like with like.
Can you use CSAT and NPS together?
Yes, and running them together is common. CSAT tells you which interactions to fix, and NPS tells you whether those fixes are changing how customers feel about the brand. Add a "why?" follow-up to both surveys, and follow up with customers who give low scores.
Bain calls the open-text follow-up "that all-important second question: Why?". It also recommends closing the loop by calling customers back (Bain, 2015).
If you run both, this checklist keeps the data usable:
Ask one rating question per survey, and keep the wording the same over time.
Use a consistent scale, and don't switch CSAT from 1–5 to 1–10 halfway through the year.
Send CSAT within hours of a resolution, while the interaction is still fresh.
Send NPS on a fixed schedule, and avoid surveying the same customer too often.
Always add an optional "What's the main reason for your score?" question.
Contact detractors and low CSAT scorers, and record what you changed as a result.
CSAT vs NPS vs CES: where does Customer Effort Score fit?
Customer Effort Score (CES) asks how easy it was to get something done, usually on a 1–5 or 1–7 scale. Like CSAT, it's a transactional metric, but it measures friction where CSAT measures feeling. The idea came from research covering more than 75,000 customers, which found that reducing effort does more for loyalty than trying to delight.
That research was published as "Stop Trying to Delight Your Customers" in Harvard Business Review (Dixon, Freeman and Toman, 2010), and covered customers using contact centres and self-service.
Here's how the three metrics compare:
CSAT: "How satisfied were you?" on a 1–5 scale, sent after a specific interaction. Best for judging agents, conversations and processes.
NPS: "How likely are you to recommend us?" on a 0–10 scale, sent periodically. Best for brand loyalty and long-term health.
CES: "How easy was it to resolve your issue?" on a 1–5 or 1–7 scale, sent after a task such as a return or a support contact. Best for finding friction in a journey.
Other support metrics worth tracking alongside them
Surveys only cover customers who reply. These operational metrics fill in the rest:
First contact resolution: the share of issues resolved in a single contact.
Contact-to-order ratio: contacts divided by orders. A ratio of 25% means one contact for every four orders. You can benchmark yours with our contact-to-order ratio calculator.
Average handle time: how long an agent spends on a ticket. Push it down too hard and satisfaction suffers.
Cost per ticket: compare it with your average order value. A £5 cost per ticket is fine for a furniture retailer with a £1,000 order value, and painful for a seller of second-hand books.
How should ecommerce support teams track CSAT and NPS?
In ecommerce support, CSAT largely tracks whether the customer got the outcome they came for, and NPS reflects how the post-purchase experience went as a whole. The way to raise both is to resolve more order, delivery and return issues. That requires your support team, human or AI, to have access to the systems that can actually fix them.
Read CSAT by outcome before you judge anyone. DigitalGenius compared how AI and human agents score on CSAT for a representative customer. Across all tickets, the AI agent averaged 8.6 out of 10 against 8.4 for human agents. When there was a positive resolution to give, the gap widened slightly, to 9.4 against around 9. When the answer was no, human agents scored comfortably higher. The outcome moved the score more than anything else, so segment CSAT by resolution type before judging agents or automation.
Give support the ability to act. A positive outcome usually means doing something: checking a carrier, reshipping an order, approving a return or issuing a refund. That's why DigitalGenius' AI agent for customer care connects to the order management system, carriers and payment providers. It investigates what went wrong, then takes the action within the brand's rules. When a person is needed, it hands the conversation to an agent with the context intact.
Treat post-purchase as an NPS driver. Delivery and returns shape whether customers come back. 71% of consumers say they are less likely to shop with a retailer again after a poor returns experience (NRF and Happy Returns, 2025). Narvar's platform data shows shoppers whose order arrived within the promised window repurchase at 42.9%, against 12.6% when it arrived late (Narvar, 2026). Across DigitalGenius customers, "where is my order?" (WISMO) queries made up around 15% of all queries in August and September 2026.
Pair automation metrics with CSAT. A deflection rate on its own can look healthy while customers leave without an answer, so report it next to CSAT.
Brands can do well on both. Scentbird automates its common WISMO, cancellation and subscription contacts, with 70% of emails getting an automatic response, and reports a CSAT above 4.9 out of 5, alongside an NPS of 68. Beauty Pie reports a 95% CSAT through automation. For more tactics, see our guide on how to increase CSAT scores.
FAQ
Is CSAT or NPS better for measuring customer service?
CSAT is better for measuring customer service, because it's tied to a specific interaction and you can act on it at the level of an agent, channel or contact reason. NPS reflects the whole brand experience, including product and price, so it's a weaker measure of service quality.
What scale does CSAT use?
CSAT most often uses a 1–5 scale. It's reported either as the percentage of 4s and 5s or as the average score. Some teams use 1–3, 1–7 or 1–10 scales, or emoji and star ratings that map onto numbers. The ACSI uses a 0–100 index. Keep whichever scale you choose consistent over time.
Can an NPS be negative?
Yes. NPS is the percentage of promoters minus the percentage of detractors, so if detractors outnumber promoters the score is negative. The lowest possible score is −100 (every respondent is a detractor) and the highest is +100 (every respondent is a promoter).
How often should you send NPS surveys?
Many brands send relational NPS surveys every three to six months, or at a set point after purchase, such as 30 days after delivery. Survey too often and response rates fall. CSAT is different: send it after each interaction, while the experience is still fresh.
What's the difference between CSAT, NPS and CES?
CSAT measures how satisfied a customer was with an interaction. NPS measures how likely they are to recommend your brand. CES measures how easy it was for them to get something done. CSAT and CES are transactional, sent after a specific touchpoint, and NPS is relational, sent periodically.
Choosing the right metric
Choose each metric by the decision it needs to inform. CSAT is for fixing interactions, CES is for removing friction, and NPS is for tracking whether customers are sticking around. Then check that the team being measured can actually change the outcome. If support can't reship, refund or rebook, no amount of coaching will move CSAT.
If you want to resolve more post-purchase tickets from start to finish, speak to the team.
