The Net Promoter Score (NPS) is the most widely adopted customer loyalty indicator in the world. Developed in 2003 by Fred Reichheld of Bain and Company in a Harvard Business Review article titled “The One Number You Need to Grow,” it rests on a single question: “How likely are you to recommend our company to a friend or colleague?” The answer, scored from 0 to 10, produces a score ranging from -100 to +100 that measures the balance between customers who actively recommend you and those who actively advise against you.
Simple to collect, easy to present in the boardroom, comparable across time and sectors: these qualities have made NPS a global standard. This guide covers the full subject: precise definition, step-by-step calculation formula, interpretation rules, France sector benchmarks for 2025-2026, survey types, best practices, and limits to understand.
The NPS measures customer loyalty and propensity to recommend a brand, product, or service. It is not an immediate satisfaction indicator like CSAT: it is a measure of the overall relationship. It evaluates the bond a customer has built with a brand over time, not their feeling after a single interaction.
The NPS question: "On a scale of 0 to 10, how likely are you to recommend [company/product/service] to a friend or colleague?"
Based on their score, respondents are classified into three categories:
NPS calculation is one of the simplest in all CX metrics. It requires no sophisticated tools, just the raw data from your survey.
NPS = % Promoters (9-10) minus % Detractors (0-6)
Passives (scores 7-8) are not included in the calculation. They serve to balance the denominator (total respondents for calculating percentages) but do not directly influence the final score.
You have collected 400 responses to your NPS survey. Here is the breakdown: 160 promoters (scores 9-10) = 40%, 120 passives (scores 7-8) = 30%, 120 detractors (scores 0-6) = 30%.
NPS = 40% minus 30% = +10
The NPS score is always expressed as a whole number without the % sign. It can range from -100 (all respondents are detractors) to +100 (all respondents are promoters). These two extremes are theoretical: in practice, observed scores range between -30 and +80 depending on the sector and the CX maturity of the organisation.
An NPS calculated on fewer than 50 responses is statistically unreliable. A variation of 2 or 3 points may be due to chance rather than a real shift in satisfaction. For contact centres handling several hundred interactions per week, systematic sampling at 20% of contacts is sufficient to ensure representativeness. Below a certain volume, it is better to measure less frequently but on a larger sample.
NPS interpretation cannot be done in absolute terms. Is a score of +30 good or bad? The answer depends on three elements: the sector, the measurement method used, and above all the trend over time.
| NPS Score | Interpretation |
|---|---|
| Below 0 | Warning signal: more detractors than promoters. Structural CX problems exist and must be addressed as a priority. |
| 0 to 20 | Acceptable but improvable. Significant room for progression, particularly on converting passives to promoters. |
| 20 to 50 | Good score. The customer relationship is solid, but can still be improved on specific friction points. |
| 50 to 70 | Excellent score. Promoters are a clear majority. The challenge is maintaining this level and activating promoters as active brand ambassadors. |
| Above 70 | Global reference level. Reserved for brands that have built a strong emotional bond with their customers. |
NPS benchmarks are predominantly built on American data (American Customer Satisfaction Index, Bain and Company). European and British consumers score structurally lower than American consumers, with an estimated gap of 3 to 5 points on average for equivalent experiences.
European adjustment to apply: consumers in France and continental Europe score structurally 3 to 5 points below American benchmarks for equivalent experiences. Do not compare against ACSI benchmarks without applying this correction. A score of +30 in Europe corresponds roughly to a score of +35 in the United States in the same sector.
The most useful comparison for a CX Director is not comparison with the market: it is comparison with yourself over time. An NPS that improves by +5 points over 12 months on your own customer base is a far more actionable signal than a score of +40 that you are not sure is good or bad for your sector.
The data below cross-references observations from Forrester Research, Bain and Company, and Armatis operational data from quality programmes deployed with our clients. Note: these benchmarks incorporate the European cultural adjustment mentioned above.
One key point: in sectors where the customer relationship is perceived as constrained (energy, telecoms, banking), a low NPS is structural, not necessarily a sign of poor management. The focus should be on actionable levers, particularly FCR and reducing customer effort, rather than targeting benchmarks from a different sector.
NPS is not a single survey. Two distinct approaches exist, responding to different objectives and complementing each other in a complete measurement system.
Relational NPS measures the customer's overall relationship with the brand at regular intervals, typically quarterly or annually. It evaluates the overall bond, independently of a recent interaction. It is the long-term trend indicator, useful for time-series comparisons and sector benchmarking. It answers the question: "How is the customer relationship doing overall?"
Transactional NPS is triggered after a specific interaction: a call to customer service, a purchase, a delivery, a complaint submission. It measures the propensity to recommend in the precise context of that interaction. It is far more granular than relational NPS and identifies the touchpoints that degrade or improve the relationship. Post-complaint NPS, for example, is often 20 to 30 points below the overall NPS, which highlights the importance of managing difficult situations for long-term loyalty.
Combining both gives a complete picture: relational NPS shows where the relationship stands, transactional NPS shows where it degrades or improves.
The quality of NPS data depends as much on when and how the survey is delivered as on the question itself. Here are the principles that separate actionable NPS from NPS that merely produces numbers.
Timing is decisive. For transactional NPS, the survey must be sent within 15 to 30 minutes of the interaction, while the experience is still fresh. Beyond a few hours, response rates drop and scores are less representative of real sentiment. For relational NPS, avoid high-solicitation periods (peak sales events, product launches) that bias responses.
Always add an open question. The score alone says nothing. The open question that follows reveals the irritants and satisfaction drivers. "What led you to give this score?" or "What could we improve?" These verbatims are the raw material of action plans. An NPS without verbatims is a thermometer with no scale.
Keep the questionnaire short. NPS loses reliability when buried in a long survey. An effective NPS questionnaire includes the NPS question, one open question, and at most two contextual questions (channel used, contact reason). Beyond that, response rates fall and respondents skew toward those with the most extreme opinions.
Measure response rate as much as the score. An NPS of +45 from 15 respondents is meaningless. A response rate above 30% is a good benchmark for post-interaction surveys. Below 15%, representativeness is insufficient for operational decisions.
NPS alone is not sufficient to manage customer experience. It is complemented by two other indicators that measure different dimensions of the relationship.
CSAT (Customer Satisfaction Score) measures immediate satisfaction after a specific interaction. It is a point-in-time thermometer: it tells you whether the interaction went well, not whether the customer is loyal. High CSAT with low NPS signals a customer who is satisfied day-to-day but would not recommend the brand overall, often because the overall image or perceived value falls short.
CES (Customer Effort Score) measures the effort the customer had to put in to complete their task. It is the best short-term churn predictor. According to Gartner, 96% of customers who experienced a high-effort interaction become less loyal, compared to only 9% of those who experienced a low-effort interaction.
A customer with high NPS, high CSAT and low CES is actually at risk: they are satisfied but exhausting themselves to interact with you. For a deeper look at combining these three indicators, our NPS, CSAT, CES guide details how to use them together for coherent CX management.
Improving NPS is not decreed: it is built on specific levers, each acting on a different dimension of the customer experience.
Reduce repeat contacts (FCR). The first unresolved contact is the main generator of detractors. A customer who has to call back twice for the same problem will not recommend you, regardless of the quality of each individual interaction. Improving First Contact Resolution is the most direct NPS lever in a contact centre. Based on Armatis operational observations, each FCR point gained translates into 0.5 to 1 point of CSAT on average, with a delayed positive effect on NPS. Our article on what makes a good CSAT score details the relationship between these indicators.
Treat detractors quickly. Detractors are not lost. Contacting a detractor customer within 24 hours of their score and offering a concrete solution regularly transforms a detractor into a passive, or even a promoter. This is "close the loop": completing the feedback cycle with a visible action. Organisations that systematically practise close-the-loop see their NPS progress by 10 to 20 points over 12 months.
Convert passives into promoters. Passives (scores 7-8) are often neglected because they do not damage the score. Yet they represent the largest reserve of improvement. Understanding why a customer does not give a 9 or 10 when they are satisfied is often more instructive than analysing detractors. Passive verbatims reveal "discreet irritants": not serious enough to create declared dissatisfaction, but sufficient to prevent enthusiasm.
Segment NPS by channel and by contact reason. An overall NPS of +30 may conceal an NPS of +50 on voice and +10 on chat. Segmenting NPS by channel, interaction type, and customer segment identifies real friction points rather than diluting actions into a generic improvement plan. For contact centres, segmentation by contact reason is particularly revealing: post-complaint NPS, post-simple-information NPS, and post-claim NPS do not follow the same dynamics and do not improve with the same levers.
NPS is among the most common contractual commitments in BPO outsourcing SLAs. But contractualising NPS without safeguards can create perverse effects: a provider that optimises its score without genuinely improving the experience, or a client that manages on a quarterly NPS without seeing weekly deterioration signals.
Best practices for integrating NPS into a BPO SLA: define a monthly minimum threshold by channel (not a global NPS), specify the shared measurement method between both parties, and provide for a double-measurement mechanism (the provider measures continuously, the client measures independently via a third party). Comparing both sets of measurements detects complacency bias and representativeness problems.
NPS must also be complemented by FCR and repeat contact rate in SLAs: these two operational indicators allow the contact centre's performance to be linked to the measured NPS, and detect deteriorations before they show up in the quarterly score. Our guide on managing an outsourced contact centre details how to build this complete measurement framework.
NPS is useful. It is not infallible. Knowing its limits allows you to use it correctly rather than asking of it what it cannot provide.
Its first limit is that it measures a declared intention, not future behaviour. A customer who gives an 8 may leave tomorrow. A customer who gives a 6 may stay for years because switching costs are high. The correlation between NPS and actual retention is real but imperfect.
Its second limit is cultural bias. Japanese, German, and French consumers score structurally lower than Americans. Comparing international NPS without cultural adjustment is like comparing temperatures in Celsius and Fahrenheit without conversion.
Its third limit is sensitivity to phrasing and timing. An NPS sent just after a successful delivery will be systematically higher than one sent after a complaint, even if both are measuring "overall loyalty." The measurement context influences the score as much as the experience itself.
Finally, NPS is a stock indicator (the state of the relationship at a point in time), not a flow indicator (what is happening in daily interactions). To manage continuous improvement, it must be complemented by transactional indicators such as CSAT and CES.
What is NPS?
The Net Promoter Score (NPS) is a customer loyalty indicator that measures the propensity of customers to recommend a brand, product, or service. It is calculated from a single question scored from 0 to 10 and produces a score ranging from -100 to +100. Developed in 2003 by Fred Reichheld of Bain and Company, it is today the most widely used loyalty indicator in the world.
How do you calculate NPS?
NPS = % of promoters (scores 9-10) minus % of detractors (scores 0-6). Passives (scores 7-8) are not included in the formula. Example: 40% promoters and 25% detractors give an NPS of +15. The score is expressed as a whole number without the % sign.
What is a good NPS score?
An NPS above 0 is positive. Between 20 and 50, it is good. Above 50, it is excellent. These benchmarks vary by sector: an NPS of +20 is strong in telecoms, average in luxury. In Europe, scores are structurally 3 to 5 points below US benchmarks for equivalent experiences. The most useful comparison remains against your own score over time.
What is the difference between relational NPS and transactional NPS?
Relational NPS measures the overall relationship at regular intervals (quarterly or annually) independently of a recent interaction. Transactional NPS is triggered after a specific interaction (call, purchase, complaint) and measures propensity to recommend in the precise context of that interaction. Combined, they give a complete view of relationship health.
What is the difference between NPS, CSAT, and CES?
NPS measures overall loyalty and propensity to recommend. CSAT measures immediate satisfaction after a specific interaction. CES measures the effort the customer had to put in to complete their task. The three indicators are complementary: NPS shows where the relationship stands, CSAT shows whether interactions go well, CES shows whether they are easy.
How do you improve NPS in a contact centre?
The most effective levers are: improving FCR (each FCR point gained translates into 0.5 to 1 CSAT point and a delayed positive effect on NPS), treating detractors within 24 hours (close the loop), segmenting NPS by channel and contact reason to identify real friction points, and converting passives (scores 7-8) by understanding what prevents them from giving a 9 or 10.
Armatis is a European specialist in customer relations and business process outsourcing (BPO), operating across multiple continents with thousands of employees serving companies of all sizes and sectors. The company designs and manages end-to-end customer service operations: multichannel contact centres, complaints handling, technical support, back-office and digitised processes. Backed by integrated technology infrastructure and the ability to adapt to any sectoral and regulatory context, Armatis helps its clients combine operational performance, quality of experience and cost control, wherever they need it.
Contact our teams to discuss your challenges and find out how we can support you
Join the leaders who trust our multilingual and technological expertise.