Customer Experience Metrics: NPS, CSAT, and CES Compared (2026)

Customer experience is measured by three universally-used metrics: Net Promoter Score (NPS), Customer Satisfaction Score (CSAT), and Customer Effort Score (CES). Each answers a different question, predicts a different business outcome, and has its own scale and formula. Picking the right one for the right touchpoint — or, more accurately, combining all three — is the difference between feedback theatre and a feedback programme that actually changes the business.

This is the definitive reference for the three metrics: definitions, formulas, worked examples, when to use which, common mistakes, and how mature CX programmes combine them. Every benchmark is sourced from published research (Bain & Company, Harvard Business Review, ACSI, Forrester CX Index, Gartner) — no proprietary platform data is used, by policy.

TL;DR — the three metrics at a glance

Metric	Question	Formula	Best for
NPS	Would you recommend us?	% Promoters − % Detractors	Long-term loyalty
CSAT	How satisfied are you?	Satisfied / Total × 100	Specific interactions
CES	How easy was it?	Low-effort / Total × 100	Friction-sensitive touchpoints

If you only deploy one metric, pick NPS for general CX measurement, CSAT for high-volume operational feedback (retail, hospitality), or CES for support and transactional flows. Mature programmes use all three — the rest of this article explains how.

---

Net Promoter Score (NPS)

What it is

NPS measures customer loyalty — specifically, the willingness of customers to recommend a company to their peers. It is the most widely-used customer experience metric, used by a majority of Fortune 1000 companies.

The formula

NPS = % Promoters − % Detractors

Respondents rate "How likely are you to recommend us?" on a 0-10 scale and are classified into three groups:

• Promoters (9-10) — loyal enthusiasts who recommend you
• Passives (7-8) — satisfied but unenthusiastic
• Detractors (0-6) — unhappy customers

Passives appear in the denominator (the total) but do not add to or subtract from the score. The result is a single number between -100 and +100.

Some platforms — including kazva.bg — use an extended 1-100 scale with proportionally adjusted cutoffs (promoters 80-100, passives 50-79, detractors below 50). The wider scale captures finer differences in sentiment that the coarser 0-10 scale would miss.

Use the free NPS calculator to compute your score, or read the step-by-step guide with multiple worked examples. The full definition lives in the NPS glossary entry.

History

NPS was introduced by Fred Reichheld of Bain & Company in the December 2003 Harvard Business Review article "The One Number You Need to Grow." Reichheld argued that the recommendation question — easier to answer and more predictive than complex multi-question surveys — correlates strongly with subsequent revenue growth. The metric was developed in collaboration with Satmetrix.

Industry benchmarks

Published industry averages (Bain & Company, Retently, ACSI):

• Hospitality and education: +35 to +55
• Retail: +30 to +45
• Financial services: +25 to +40
• Airlines: +20 to +35
• Telecommunications: +15 to +30
• Municipal services: +15 to +30
• Public transport: +10 to +25

Always compare your NPS against your own sector, not against an absolute target. A +25 in public transport is roughly equivalent to a +50 in hospitality.

When NPS works

• Brand-level loyalty tracking (board reporting, year-on-year comparison)
• Pre/post product launches
• Strategic CX initiatives where the success metric is "would they come back / recommend us"
• B2B account-level relationship measurement (with the variant "rNPS" for relationship NPS)

When NPS fails

• Evaluating specific touchpoints (the question is too high-level — use CSAT instead)
• Diagnosing operational issues (use CSAT segmented by location or CES on support)
• Comparing product variants (use CSAT or A/B testing)
• Small samples (NPS is statistically noisy below 100 responses)

---

Customer Satisfaction Score (CSAT)

What it is

CSAT captures the percentage of customers who were satisfied with a specific interaction — a purchase, visit, service experience, or support contact. Unlike NPS, which measures long-term loyalty, CSAT is a snapshot of a particular touchpoint.

The formula

CSAT % = (Number of Satisfied Responses ÷ Total Responses) × 100

The result is a percentage between 0 and 100, never negative. The denominator includes neutral and dissatisfied responses — omitting them inflates the score artificially.

Classical 1-5 scale classification:

• Satisfied: 4-5
• Neutral: 3
• Dissatisfied: 1-2

The 1-100 kazva.bg variant maps satisfied = 80-100, neutral = 50-79, dissatisfied = below 50.

Compute your CSAT with the free CSAT calculator, or read the step-by-step guide. Full definition: CSAT glossary entry.

History

CSAT predates NPS by decades. Variants of the satisfaction question have been in market research since the 1970s. The American Customer Satisfaction Index (ACSI), launched in 1994 at the University of Michigan's Ross School of Business, formalised the methodology for cross-industry benchmarking. The European Customer Satisfaction Index (ECSI) followed in 1999.

Industry benchmarks

Published industry baselines (ACSI, Forrester CX Index):

• Retail and hospitality: 75-85%
• Financial services: 70-80%
• Telecommunications: 65-75%
• Public services: 55-70%
• Public transport: 60-70%

A 65% CSAT in public services is roughly equivalent to an 80% in retail — sector context matters.

When CSAT works

• Operational management — comparing locations, shifts, counters, staff members
• High-volume transactional environments (retail, hospitality, public services)
• Quick feedback loops where the trend matters more than the absolute number
• Immediate quality control — a CSAT drop within a single shift signals real operational problems

When CSAT fails

• Measuring long-term loyalty (CSAT customers can still defect — use NPS)
• Diagnosing process friction (use CES)
• Triggered only after "good" interactions (selection bias inflates the score)
• Sent days after the interaction (memory fades, the score regresses to the mean)

---

Customer Effort Score (CES)

What it is

CES measures how much effort a customer had to spend to accomplish their goal — resolving a problem, completing a purchase, finding information. Unlike CSAT (satisfaction) or NPS (loyalty), CES focuses specifically on friction. Research shows that effort predicts churn better than satisfaction does, particularly in support and transactional contexts.

The formula

Modern CES uses a percentage form:

CES % = (Low-Effort Responses ÷ Total Responses) × 100

The classical 1-7 agreement scale ("the company made it easy for me to handle my issue"):

• Low effort: 5-7 (agree, somewhat agree, strongly agree)
• Neutral: 4
• High effort: 1-3

The kazva.bg 1-100 variant: low effort 80-100, neutral 50-79, high effort below 50.

Older implementations used the average score (1-5 or 1-7) instead of the percentage form. Most modern platforms have moved to the percentage form because it is easier to communicate and directly comparable to CSAT.

Compute your CES with the free CES calculator. Full definition: CES glossary entry.

History

CES was introduced in 2010 by Matthew Dixon, Karen Freeman, and Nicholas Toman of CEB (now Gartner) in the Harvard Business Review article "Stop Trying to Delight Your Customers." Based on more than 75,000 customer interactions, the research found that the most reliable predictor of customer loyalty was not satisfaction or perceived delight — it was the effort customers had to expend. Customers who spent high effort to resolve issues were 4× more likely to defect, even when satisfied with the outcome.

Industry benchmarks

Published industry baselines (Gartner, Harvard Business Review, Forrester):

• SaaS support: 65-80%
• Financial services: 60-75%
• Telecommunications: 55-70%
• Public services: 50-65%

When CES works

• Support contacts (phone, chat, ticket, email)
• Checkout and onboarding flows
• Returns, refunds, and complaint processes
• Password resets, account management, self-service
• Any transactional moment where friction predicts churn

When CES fails

• Measuring overall brand sentiment (effort doesn't apply to passive brand affinity — use NPS)
• Casual interactions where effort isn't a meaningful concept (use CSAT instead)
• Comparing products (use CSAT or A/B testing)

---

Side-by-side comparison matrix

Dimension	NPS	CSAT	CES
Question	Would you recommend us?	How satisfied are you?	How easy was it?
Scale	0-10 (or 1-100)	1-5 (or 1-100)	1-7 agreement (or 1-100)
Range	-100 to +100	0% to 100%	0% to 100%
Time horizon	Long-term loyalty	Immediate satisfaction	Single interaction
Predicts	Revenue growth, retention	Operational quality	Churn after friction events
Best for	Brand health, B2B accounts	Retail, hospitality, public services	Support, onboarding, returns
Year introduced	2003 (Reichheld / Bain)	1970s; formalised 1994 (ACSI)	2010 (CEB / Gartner)
Survey length	1 question	1 question	1 question
Minimum sample	100+ (noise sensitive)	50+ (more forgiving)	50+ per touchpoint

How mature CX programmes combine all three

The metrics are complementary, not competing. A mature CX programme runs each at the level it was designed for:

Strategic layer — NPS quarterly

NPS measured quarterly at the relationship level. One number per major customer segment, reported to the board. Used to track the answer to "is our overall customer health improving?" — a slow, strategic metric.

Operational layer — CSAT continuous

CSAT collected continuously at every customer touchpoint that has volume — retail checkouts, restaurant tables, service counters, support contacts, web checkouts. Aggregated per location, per shift, per staff member. Used by frontline managers to spot operational drift within hours.

Process layer — CES at friction-prone touchpoints

CES specifically deployed at touchpoints where friction is the dominant variable — support tickets, returns, password resets, complex onboarding flows. Used by process owners to identify where customers are bouncing or escalating.

The layered approach in practice

A bank running all three might see this pattern in a typical quarter:

• NPS at +35, stable quarter-on-quarter. The board sees the brand-health story is steady.
• CSAT at 78% overall, but 62% at one specific branch. The regional manager investigates and finds a staffing gap.
• CES at 71% for password resets, but 42% for the loan-application flow. Process owners prioritise the loan flow for redesign.

Each metric drove a different decision at a different level of the organisation. None of the three would have surfaced the full picture on its own.

Common mistakes when picking metrics

Mistake 1 — picking NPS because everyone else uses it

NPS is the most-cited metric in CX literature, so teams default to it. But for high-volume operational environments (retail, hospitality, public services), CSAT segmented by location produces actionable data within hours; NPS produces directional data within quarters. Pick the metric for the decision cycle, not the metric for the marketing deck.

Mistake 2 — running CES on irrelevant touchpoints

CES asks "how easy was it?" — a question that makes sense after a support contact, a checkout, a return. It makes less sense after a casual store visit ("how easy was browsing?") or a brand interaction ("how easy was reading our ad?"). Effort isn't a meaningful concept everywhere.

Mistake 3 — comparing scores across metrics

A 78% CSAT is not directly comparable to a 78% CES is not comparable to a +35 NPS. They measure different things on different scales with different distributions. Trend each metric against its own history; compare across organisations within the same metric only.

Mistake 4 — collecting metrics without closing the loop

The biggest failure mode in any CX programme is collecting feedback without responding to it. A customer who scored 2/10 and never heard back is more likely to churn than one who was never asked. Always have a closed-loop process: low scores trigger an investigation, an action, and ideally a follow-up to the respondent.

Mistake 5 — sending the survey too late

All three metrics regress to the mean as the gap between interaction and survey grows. CSAT and CES both work best within hours of the interaction; NPS tolerates a day or two but degrades after that. Survey late, get noise.

EU and Bulgarian regulatory context

Customer feedback measurement in the EU sits under several regulatory frameworks that influence the methodology:

ISO 9001 §9.1.2 — Customer satisfaction monitoring

The international quality management standard requires organisations to "monitor customers' perceptions of the degree to which their needs and expectations have been fulfilled." Most certifying bodies (TÜV, SGS, Bureau Veritas) accept CSAT or NPS as evidence, provided the methodology is documented and the data is dated, segmented, and acted on.

CAF 2020 (Common Assessment Framework)

The EU's self-assessment framework for public administration. Sub-criterion 6.1 ("citizen-oriented results") explicitly expects continuous citizen feedback with documented evidence. CSAT or NPS data segmented by service centre, time period, and language meets the requirement.

ISO 45001 §5.4 — Worker participation

For employee feedback, ISO 45001 requires documented evidence of consultation and participation, particularly in health-and-safety decisions. Employee NPS (eNPS) and employee CSAT measured continuously are common evidence formats — paper-based surveys are increasingly insufficient because they lack the timestamp and anonymity guarantees auditors expect. See our ISO 45001 worker participation guide for details.

GDPR considerations

Anonymous-by-default feedback collection is the safest GDPR posture. Avoid collecting names, email addresses, or other PII unless the closed-loop response process specifically requires it (and even then, treat the identifier as a separate, deletable record). QR-based feedback collection — no login, no registration — is the lowest-risk pattern.

Frequently asked questions

Which metric should I start with?

If you have no existing measurement, start with CSAT. It is the easiest to interpret, the most operational, and the most forgiving statistically. Add NPS once you have an executive who wants a single number for the board pack. Add CES once you have an identifiable friction-prone touchpoint to monitor.

Can I use all three on the same survey?

Yes, and many enterprise CX programmes do. But each adds friction — the longer the survey, the lower the response rate. For a one-question survey, response rates of 20-40% are achievable. For three questions, expect 10-25%. Pick based on your data needs vs participation tradeoff.

How often should I measure?

NPS — quarterly at the strategic level, monthly at the relationship level for B2B accounts. CSAT — continuously, with daily or weekly reporting. CES — triggered by specific events (every support contact, every checkout, every return), not on a fixed cadence.

What if my industry isn't in the benchmark tables?

Use the closest analogue and benchmark against your own historical baseline. A first-year score of any kind is just a baseline; the second-year delta is what matters. Avoid comparing across industries — sector dynamics are too different.

How do I avoid bias in the scores?

Three guardrails: (1) survey every customer or a random sample of every customer, not just the ones who completed the journey successfully; (2) keep responses anonymous; (3) report sample size alongside the score. If any of these is missing, the score should be treated as directional only.

Should I use the 0-10 / 1-5 / 1-7 scale or the 1-100 scale?

For external benchmarking against published industry data, use the classical scales (0-10 for NPS, 1-5 for CSAT, 1-7 for CES). For internal trend analysis where sensitivity to small changes matters, use the 1-100 scale. The kazva.bg platform supports both and converts between them when needed.

How does CES differ from "delight"?

Older CX theory assumed that delighting customers — exceeding their expectations — was the path to loyalty. The 2010 CES research found this was wrong. Most customers don't want to be delighted; they want their problem to go away. Removing friction is much more leverage than adding delight, especially in service-heavy industries.

What is rNPS / eNPS / pNPS?

Variants of NPS targeting different audiences. rNPS (relationship NPS) measures the overall account relationship in B2B. eNPS (employee NPS) measures employee willingness to recommend the employer. pNPS (product NPS) measures recommendation for a specific product rather than the whole company. All three use the same formula but reframe the question.

Use the calculators

All three calculators are free, browser-side, no signup:

• NPS Calculator — with industry benchmarks and interpretation guide
• CSAT Calculator — same pattern, customer satisfaction variant
• CES Calculator — completes the triad

For step-by-step guides:

• How to calculate NPS step-by-step
• How to calculate CSAT step-by-step

For terminology references:

• NPS glossary entry
• CSAT glossary entry
• CES glossary entry
• Full CX glossary

From measurement to action

Measuring NPS, CSAT, and CES is the easy part. The hard part is converting scores into operational decisions. A mature feedback programme has three components beyond the metrics:

• Continuous collection — feedback flowing 24/7, not annual surveys. QR codes at physical touchpoints, post-interaction email/SMS, in-app prompts.
• Real-time alerting — low scores route to the responsible manager within minutes, before the customer churns or escalates publicly.
• Closed-loop response — every low score gets an investigation, an action, and a follow-up to the respondent. This is the loop that converts dissatisfied customers into retained ones.

kazva.bg automates all three across QR codes, web, email, and SMS. Per-location segmentation, real-time alerts, AI-generated weekly reports. See how the platform works, or book a 25-minute consultation.

References

• Reichheld, F. (2003). "The One Number You Need to Grow." Harvard Business Review, December 2003.
• Reichheld, F. and Markey, R. (2011). The Ultimate Question 2.0. Harvard Business Review Press.
• Dixon, M., Freeman, K., and Toman, N. (2010). "Stop Trying to Delight Your Customers." Harvard Business Review, July-August 2010.
• Dixon, M., Toman, N., and DeLisi, R. (2013). The Effortless Experience. Portfolio.
• American Customer Satisfaction Index (ACSI). University of Michigan Ross School of Business.
• European Customer Satisfaction Index (ECSI) framework.
• Forrester Customer Experience Index — annual publications.
• Gartner / CEB Customer Effort Score research.
• Bain & Company. "Net Promoter System" research publications.