Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Use CSAT when you need to know how a customer felt about a specific interaction, product, or service. Use NPS when you need to know whether customers would recommend the organization as a whole. The two metrics answer different questions, so the right choice depends on which decision your team can actually act on. Running both is reasonable only when each one feeds a separate decision.
What CSAT measures
CSAT asks how satisfied a customer was with a particular product, service, or interaction. The metric name does not fix a scale or a cutoff, so a CSAT number is only readable if the survey states its question, its response scale, and which answers count as satisfied.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
How to Create an Effective Advisory Board (Small Business Entrepreneur Tool Kit Book 1) | $2.99 | Buy on Amazon |
SurveyMonkey’s June 2026 comparison of CX metrics gives one common convention: on a 1–5 scale, count the ratings of 4 and 5, divide by the total number of responses, and multiply by 100. That is an example of one convention, not a universal formula. A team that counts only 5s, or that uses a 1–10 scale with a different cutoff, produces a number that cannot be compared directly with this one.
Illustration (hypothetical numbers): if 200 customers rate a support ticket on a 1–5 scale and 150 of them answer 4 or 5, the CSAT is 150 ÷ 200 × 100 = 75% under that convention. The figure says nothing about the 50 customers who gave a 3 or lower, which is why the distribution of answers is worth checking alongside the headline percentage.
#1 Best Overall
What NPS measures
NPS is built on a single question. Bain & Company gives its standard wording as “How likely are you to recommend us to a friend or colleague?” Respondents choose a whole number from 0 to 10. Scores of 9 and 10 are promoters, 7 and 8 are passives, and 0 through 6 are detractors.
The score is the percentage of promoters minus the percentage of detractors. Passives are part of the total base but do not appear in the subtraction. Because the result is a difference between two percentages, it runs from −100 to +100.
Illustration (hypothetical numbers): from 100 responses, 50 promoters, 30 passives, and 20 detractors give an NPS of 50 − 20 = +30. The same 100 people could also be described as 50% promoters and 20% detractors; the score simply compresses those two shares into one number.
Recommended Free Tools
Bain also recommends stating which type of NPS is being measured: after a single experience or journey, at relationship or brand level, or competitively. A score collected after a support call and a score collected across the whole customer relationship use the same formula but measure different things.
Side-by-side comparison
| Decision axis | CSAT | NPS |
|---|---|---|
| Main question | How satisfied were you with this product, service, or interaction? | How likely are you to recommend us to a friend or colleague? |
| Scope | A specific product, transaction, or service encounter | Brand or customer relationship; can also be collected after a defined journey |
| Typical timing | Soon after the interaction, while details are fresh | Periodically for relationship sentiment, or at a clearly defined journey point |
| Useful action | Find and fix a specific experience or service problem | Track advocacy or loyalty signals and investigate the reasons behind them |
| Scoring | Depends on the chosen scale and satisfied-response convention (for example, share of 4s and 5s on a 1–5 scale) | Percentage of promoters (9–10) minus percentage of detractors (0–6) on a 0–10 scale |
| Main interpretation risk | A figure is misleading if the scale or satisfied cutoff is not disclosed | A single recommendation answer is not the whole relationship, and a higher score is not an end in itself |
Sources for the table: Bain & Company’s NPS methodology and score guidance, and SurveyMonkey’s June 2026 metric comparison.
Matching the metric to a decision
Post-interaction feedback: use CSAT
Suppose your support team wants to know whether a ticket was resolved to the customer’s satisfaction. Send a short CSAT question when the ticket is closed, using the same 1–5 scale every time and the same satisfied cutoff in every report. The result points to a queue, a script, or a handoff step that can be changed this week. A high CSAT on tickets does not tell you whether those customers would recommend the company, so do not use it for that purpose.
Relationship-level sentiment: use NPS
Suppose leadership wants a read on how customers feel about the company as a whole. Survey a defined group of active customers on a fixed schedule, such as quarterly, using the standard 0–10 recommendation question and one open-text follow-up asking for the main reason. Read the score alongside retention and renewal data. A movement in NPS is a prompt to investigate, not a diagnosis.
Using both without duplicating requests
Both metrics can serve the same organization when each feeds a different decision. CSAT typically drives short-cycle operational fixes, while NPS informs periodic reviews of the relationship. Keep the two surveys from landing on the same customer in the same week, and make sure each report names the metric it uses, its scale, and its sample.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the evidence supports, and where it stops
- Bain’s own case history. Bain & Company’s history of the metric states that likelihood to recommend was the strongest behavior-predicting question in 11 of 14 industry case studies. In two of the remaining three, another question won, with likelihood to recommend close behind. Bain does not state the year of these studies on that page. This is the originator’s account of its own research, not independent proof that NPS predicts growth in every organization.
- A published critique. Nicholas I. Fisher and Raymond E. Kordupleski’s 2018 arXiv preprint, “What is Wrong with Net Promoter Score,” argues that NPS has weaknesses as a high-level market research measure. The assessment here is based on the preprint’s abstract, and it shows that the metric’s validity is contested rather than settling the comparison.
- Bain’s warning against score chasing. Fred Reichheld, a Bain Fellow, wrote in a Bain-published article: “The goal of the Net Promoter System® is not to increase your Net Promoter Score.” He argues that a survey answer at one moment may not capture the whole relationship, and that direct feedback should be combined with operational and behavioral signals.
- No universal “good score.” No independently verified cross-industry benchmark for either metric has been established. Acceptable CSAT and NPS levels vary by industry and by how the survey was collected, so a target copied from another company is not a reliable standard.
What to pair with either score
- The sample: who was surveyed, how many responded, and whether the group changed between periods.
- An open-text reason, which is the fastest way to learn why a score moved.
- Operational data for the same period, such as repeat purchases, renewals, or repeat service contacts.
- A consistent collection method, so that a change in the number reflects a change in customers rather than in the survey.
A score on its own does not identify the cause of a problem or prove loyalty. Read it as a signal that points to where to look next.
