What Is a Good NPS Score? Benchmarks and How to Improve It

Net Promoter Score (NPS) is one of the most widely used loyalty metrics in business, yet the question that trips up most teams is deceptively simple: what actually counts as a good score? A number that looks disappointing in one industry can be best-in-class in another, and a single headline figure tells you little without context. This guide explains exactly how NPS is calculated, what realistic benchmarks look like, and the concrete steps that move the number in the right direction.

Table of contents

• How NPS is actually calculated
• So what is a "good" NPS score?
• Benchmarks by industry
• Why raw benchmarks can mislead you
• How to improve your NPS
• Common NPS mistakes to avoid
• Frequently Asked Questions

How NPS is actually calculated

NPS comes from a single question: "How likely are you to recommend us to a friend or colleague?" answered on a 0 to 10 scale. Respondents are grouped into three buckets based on their answer:

• Promoters (9-10): loyal enthusiasts who keep buying and refer others.
• Passives (7-8): satisfied but unenthusiastic; vulnerable to competitors.
• Detractors (0-6): unhappy customers who can damage your reputation through negative word of mouth.

The formula is simply % Promoters minus % Detractors. Passives count toward the total number of responses but are excluded from the subtraction. For example, if 60% of respondents are promoters, 25% are passives, and 15% are detractors, your NPS is 60 minus 15, which equals 45. Because it is a difference of two percentages, the score ranges from -100 (everyone is a detractor) to +100 (everyone is a promoter). It is expressed as a whole number, never as a percentage with a sign. If you want a ready structure for collecting these responses, see our NPS survey type or start from the NPS survey template.

So what is a "good" NPS score?

As a rough rule of thumb, any score above zero means you have more promoters than detractors, which is a positive signal. Many practitioners use these loose tiers:

• Above 0: generally acceptable; you are net positive.
• Above 30: good; customers are clearly satisfied.
• Above 50: excellent; strong loyalty and advocacy.
• Above 70: world-class; rare and usually reserved for beloved brands.

These tiers are a useful starting point, but treat them as guardrails rather than gospel. The score that matters most is your own score measured consistently over time. A business moving from 20 to 35 is improving customer loyalty in a real, measurable way, even though 35 sits in the "good" tier rather than "excellent." Direction and trend almost always beat the absolute number.

Benchmarks by industry

Average NPS varies enormously by sector because customer expectations, switching costs, and the emotional nature of the purchase all differ. A few patterns hold reliably:

• Subscription software and consumer tech often post relatively high scores because users who stay are usually those who genuinely like the product.
• Retail and hospitality can score well when the experience is personal and memorable, which is why restaurants and hotels invest heavily in service.
• Financial services, insurance, and utilities typically score lower. These are often low-engagement, obligation-driven relationships where customers rarely feel enthusiasm even when nothing goes wrong.
• Healthcare sits in the middle, shaped heavily by wait times, communication, and perceived empathy.

The practical implication is that a bank scoring 35 may be outperforming its peers, while a SaaS product at 35 might be lagging. Always compare like with like. Industry-specific structures help here, such as our surveys for banks or surveys for restaurants, which frame the recommend question in the right context.

Why raw benchmarks can mislead you

Published benchmarks are tempting because they give you a tidy target, but several factors quietly distort them:

• Cultural response bias: respondents in some regions are far more reluctant to give top marks. A 9 in one market is a 7 in another for the identical experience.
• Survey timing: asking immediately after a delightful moment yields very different scores than asking after a billing cycle or a support escalation.
• Sample composition: if you only survey customers who recently contacted support, you skew toward people with active problems.
• Relational vs transactional NPS: a relationship survey sent twice a year measures overall sentiment, while a transactional survey after a specific interaction measures that touchpoint. The two are not comparable.

Because of all this, the most reliable benchmark is your own historical data collected with a fixed methodology. Lock down your question wording, scale, timing, and audience, then watch the trend.

How to improve your NPS

Improving NPS is not about the survey; it is about what you do with the answers. The highest-leverage move is the follow-up question immediately after the score: "What is the main reason for your rating?" This open-text response is where the value lives, because it tells you precisely why detractors are unhappy.

From there, a structured improvement loop looks like this:

• Close the loop with detractors fast. Reach out to anyone who scored 0-6 within a day or two. Many can be recovered, and the act of listening itself raises future scores.
• Find the systemic themes. Tag open-text comments to spot recurring drivers such as slow delivery, confusing pricing, or unhelpful support. Fix the cause, not the symptom.
• Convert passives, not just detractors. Passives are often one small improvement away from becoming promoters, and they are usually cheaper to win over.
• Pair NPS with a satisfaction measure. Running a CSAT survey alongside NPS shows whether dissatisfaction is broad or concentrated in specific journeys.
• Make the loop visible internally. Share verbatim comments with the teams that own each touchpoint so the feedback drives real change rather than sitting in a dashboard.

Done consistently, this turns NPS from a vanity number into an operating system for customer experience.

Common NPS mistakes to avoid

Even experienced teams undermine their own program. Watch for these traps:

• Gaming the score: begging customers for 10s or only surveying happy users produces a flattering but useless number.
• Surveying too often: fatigue tanks response rates and biases your sample toward the most opinionated customers.
• Ignoring the comment field: the number tells you that something is wrong; the comments tell you what.
• Treating one quarter as a verdict: small samples swing wildly. Look at rolling averages and statistically meaningful volumes before declaring victory or panic.

When you are ready to measure deeper satisfaction across a full journey, a broader customer satisfaction survey complements your NPS program well.

Frequently Asked Questions

Is a negative NPS always bad?

A negative score means you have more detractors than promoters, which signals a real problem worth addressing. That said, some industries with low engagement routinely sit near or below zero, so compare against sector peers and, most importantly, against your own past results before drawing conclusions.

How many responses do I need for a reliable NPS?

There is no universal minimum, but very small samples produce volatile scores that bounce on a handful of responses. For most businesses, a few hundred responses per period gives a reasonably stable read. The key is consistency: measure the same way each time so the trend is trustworthy.

What is the difference between transactional and relational NPS?

Transactional NPS is collected right after a specific interaction, such as a purchase or support call, and measures that touchpoint. Relational NPS is collected periodically and measures overall sentiment toward your brand. Keep them separate, because mixing the two distorts both.

How often should I send an NPS survey?

For relational NPS, twice a year to quarterly is common, enough to spot trends without fatiguing customers. Transactional NPS can be triggered after relevant events, but cap how often any single customer is asked so you do not erode response quality.