Net Promoter Score (NPS) measures customer loyalty with one question: "How likely are you to recommend us to friends or family?" on a scale of 0 to 10. Your score is the percentage of promoters (9 or 10) minus the percentage of detractors (0 to 6), giving a number between -100 and +100.
Key takeaways
- NPS divides respondents into promoters (9-10), passives (7-8) and detractors (0-6).
- The score is % promoters minus % detractors; passives count in the denominator only.
- Bain, who created the metric, call anything above 0 good, above 20 favourable, above 50 excellent and above 80 world class.
- The median NPS across industries in recent benchmark studies is around 40; retail averages around 45.
- The score signals loyalty. The "why" behind it is where the improvement work starts.
What is Net Promoter Score?
Net Promoter Score is a loyalty metric developed by Fred Reichheld and Bain & Company in 2003, built on the finding that willingness to recommend correlates with repurchase and growth. One question, one scale, three groups. Promoters actively bring you customers. Passives are satisfied enough to stay but not to talk about you. Detractors are at churn risk and can damage your reputation through word of mouth and public reviews. Because it is short and comparable across companies, NPS became the default boardroom CX number.
How do you calculate your NPS?
The NPS formula is: % promoters minus % detractors. Say 1,000 customers respond: 580 score 9 or 10, 320 score 7 or 8, and 100 score 0 to 6. That is 58% promoters and 10% detractors, so your NPS is 48. Passives dilute both percentages but are never subtracted, which produces a counterintuitive effect worth knowing: converting a passive into a promoter raises your score exactly as much as converting a detractor into a passive, yet the two moves require completely different work. Report the three group shares alongside the net number; the score is always a whole number, never a percentage.
What is a good NPS score in 2026?
A good NPS is anything above 0 by Bain's original interpretation: above 0 means promoters outnumber detractors, above 20 is favourable, above 50 is excellent, above 80 is world class. In practice, 2025 cross-industry benchmarks put the median around 40, with B2C brands averaging 49 against 38 for B2B, retail around 45 and consumer electronics in the mid 50s. Two cautions when benchmarking. First, survey channel and timing move scores by double digits, so external comparisons are rough at best. Second, the most useful benchmark is your own previous score: a good NPS is a better score than last quarter's, achieved without changing how you measure. The full benchmark discussion lives in what is a good NPS score.
Which NPS question should you ask, and what else?
The core NPS question is fixed: "On a scale from 0 to 10, how likely are you to recommend [company] to friends, family or colleagues?" What separates useful programmes from score theatre is the follow-up: "What is the most important reason for your score?" That open answer turns a number into a diagnosis. Keep the survey at those two questions; each additional question costs response rate. Why two is the right number, and when more is defensible, is covered in how many questions an NPS survey should have and in the four-leaf clover of effective surveys.
What is eNPS?
Employee Net Promoter Score (eNPS) applies the same mechanic internally: "How likely are you to recommend us as an employer?" It matters for customer experience because the two travel together; disengaged teams rarely produce enthusiastic customers, and Forrester explicitly named declining employee experience among the drags on its falling CX Index. Measure eNPS with the same discipline: anonymously, regularly, and with an open follow-up question. Expect structurally lower numbers than customer NPS (employees grade harder; anything above 10 to 30 is commonly considered healthy) and resist comparing the two directly. The practical link to CX runs through coaching: teams that see real customer feedback about their own work, including the praise, tend to move both scores at once.
Why is NPS not enough on its own?
NPS on its own tells you that loyalty moved, never why it moved. Three additions turn it into a working system. Pair it with transactional metrics: CSAT and CES locate which touchpoint created the promoter or the detractor. Analyse the open answers at scale: AI analysis such as ISAAC from Hello Customer classifies every verbatim by topic and sentiment, and key driver analysis quantifies which topics actually move your score. And act on detractors fast: automated close-the-loop workflows get a human response to unhappy customers within days, which is where retention is won; the playbook is in how to handle NPS detractors. For the strategic frame around all this, see Forrester's four levels of CX maturity.
What are the criticisms of NPS?
NPS attracts serious criticism, and a serious programme should know it. The score is easy to game: surveying only happy moments, prompting for 9s at the counter, or quietly excluding complainers inflates the number while actual loyalty stays put, which is why compensation should never hang on the raw score. The single number hides distribution: an NPS of 40 built from 50% promoters and 10% detractors describes a calmer business than the same 40 built from 60% promoters and 20% detractors. Cultural response styles differ by market, making cross-country comparison treacherous; a reserved market's 8 often carries the enthusiasm of another market's 10. And the metric explains nothing by itself. All four criticisms share a root: they apply to NPS used as a trophy. Used as the entry point to driver analysis and detractor follow-up, the weaknesses shrink, because decisions rest on the feedback behind the number rather than on the number.
How do companies use NPS in practice?
In practice, mature programmes run NPS at two levels with different rhythms. Relationship NPS goes to a rolling sample of the whole customer base, continuously or quarterly, and feeds board reporting: the score, its trend, and the three themes currently driving it. Transactional NPS (or better, CSAT and CES) follows key journeys and feeds operational teams. The cuts matter more than the total: per segment, per country, per store or product line, because a stable overall 40 routinely hides a 55 in one segment and a 15 in another. Realistic target-setting follows from measurement discipline: on an unchanged methodology, moving a mature score by 5 to 10 points in a year is ambitious, and any faster jump deserves suspicion before celebration. The teams that get lasting value pair every score movement with its explanation and log improvement initiatives against later score changes through impact tracking.
How do you start an NPS programme in 90 days?
A workable NPS launch fits in a quarter. Weeks one to four: fix the methodology (question wording, channel, quarantine rules, sampling plan) and wire the survey to your customer data so sends trigger automatically; decisions made here are permanent, since changing them later breaks your trend. Weeks five to eight: run the first waves, resist reading the absolute score too seriously, and stand up the two workflows that create value: detractor routing with a 48-hour response target and weekly theme review of the open answers. Weeks nine to thirteen: report the first full picture (score, distribution, top three drivers), pick one driver to fix, and log the fix so the next quarter's movement has a candidate explanation. What you should not do in the first 90 days: benchmark against industry tables, set score targets, or attach the number to anyone's bonus. The programme's first job is a trustworthy baseline; everything else builds on it.
NPS at a glance
| Score range | Reading |
|---|---|
| Below 0 | More detractors than promoters; churn risk is structural |
| 0 to 20 | Good; loyalty base exists but is fragile |
| 20 to 50 | Favourable; clear promoter surplus |
| 50 to 80 | Excellent; loyalty is a growth engine |
| Above 80 | World class; rare outside cult brands |
FAQ about Net Promoter Score
What question does NPS use?
"On a scale from 0 to 10, how likely are you to recommend us to friends, family or colleagues?" followed ideally by an open question asking the main reason for the score.
Who counts as a promoter, passive and detractor?
Promoters answer 9 or 10, passives 7 or 8, detractors 0 through 6. Only promoters and detractors enter the formula; passives affect the percentages through the total count.
Can NPS be negative?
Yes. NPS ranges from -100 (everyone is a detractor) to +100 (everyone is a promoter). A negative score means detractors outnumber promoters.
How often should you measure NPS?
Most mid-market companies measure relationship NPS continuously on a rolling sample or quarterly, plus transactional NPS after key journeys. Continuous measurement smooths seasonal distortion.
Is NPS still relevant in 2026?
Yes, as a compass rather than a steering wheel. Its value in 2026 comes from the open feedback attached to it and the speed with which teams close the loop on detractors, more than from the number itself.
What is the difference between relationship and transactional NPS?
Relationship NPS surveys a rolling sample about the brand as a whole and tracks strategic loyalty. Transactional NPS follows a specific interaction; most teams are better served using CSAT or CES there and reserving NPS for the relationship level.
Can you survey the same customers repeatedly for NPS?
Yes, with a quarantine period of at least a quarter between touches. Rolling sampling spreads the load across the base, keeps the trend continuous and avoids training customers to ignore your surveys.
---