Net Promoter Score (NPS): What It Is, How to Calculate It and What a Good Score Is

NPS is one of the most widely used customer experience metrics because it is simple: one question, one number, easy to track over time. That simplicity is also its weakness. The number tells you that loyalty moved, not why, and an average across all customers can hide very different stories in different segments.

This guide covers how NPS works, how to calculate it with worked examples, what a good score is, which questions to ask alongside it, how to analyse the comments, and how NPS compares with CSAT and CES.

What is Net Promoter Score?

Net Promoter Score was introduced by Fred Reichheld of Bain & Company in 2003 as a simple measure of customer loyalty. It rests on one idea: customers who would recommend you are more likely to stay, buy more and bring others, and customers who would not are more likely to leave and to talk you down.

The NPS question is:

> "On a scale from 0 to 10, how likely are you to recommend [company / product] to a friend or colleague?"

Responses fall into three groups:

Passives count towards the total number of responses but not towards the score itself.

How to calculate NPS

NPS = (number of promoters ÷ total responses × 100) − (number of detractors ÷ total responses × 100)

Or more simply: % promoters − % detractors.

Worked example 1

You receive 200 responses:

NPS = 55 − 20 = +35.

Worked example 2: why the average hides the story

Now split the same 200 responses by channel:

The overall +35 looks healthy. The in-store experience is close to zero. Acting on the average would mean acting on neither group. Always break NPS down by the segments that matter to your business: channel, customer tier, region, product, tenure.

NPS in a spreadsheet

If scores are in column A, from A2 to A201:

Note that NPS is not the average score. An average of 8 out of 10 can produce very different NPS results depending on how the answers are spread.

What is a good NPS score?

NPS ranges from −100 (every respondent is a detractor) to +100 (every respondent is a promoter). As a general guide, Bain & Company, which created the metric, describes a score above 0 as good, above 20 as favourable, above 50 as excellent and above 80 as world class.

Treat these as rough orientation, not targets. What counts as good depends a lot on:

• Industry. Scores vary widely between sectors such as telecom, banking, retail and software, so compare with peers, not with a universal number.
• Survey method. Relationship surveys sent to all customers usually score differently from transactional surveys sent after a support case or a delivery.
• Who you ask and when. Surveying only recent buyers, or only after good experiences, inflates the score.
• Region and culture. People in some countries rarely give 9s and 10s, even when they are happy.

The most useful comparison is with yourself: the trend over time, and the gap between your segments.

Relationship vs transactional NPS

Many companies run both: relationship NPS for the big picture, transactional surveys to find out where it is won and lost.

NPS survey questions

The score question is only half of an NPS survey. The follow-up is where you learn what to do.

The core question

• How likely are you to recommend [company] to a friend or colleague? (0–10)

The essential follow-up (open-ended)

• What is the main reason for your score?

Optional follow-ups by group

• Promoters: What do you value most about [company]?
• Passives: What is one thing we could do to make you more likely to recommend us?
• Detractors: What disappointed you, and what could we do to fix it?

Optional context questions (keep them few)

• Which product or service is your answer mainly about?
• Would you be willing to talk to us about your answer? (with permission to follow up)

NPS survey best practices

• Keep it short. One score, one open question. Add more only if you will act on the answers.
• Time it well. Send transactional surveys within a day or two of the experience; send relationship surveys on a fixed schedule so trends are comparable.
• Use the channel your customers use: email, in-app, website, SMS, WhatsApp or a short phone call.
• Don't coach the answer. Avoid messages such as "Please give us a 10".
• Survey a representative sample, not only your happiest customers.
• Close the loop. Follow up with detractors quickly and tell customers what changed.

How to analyse NPS comments

The score tells you whether loyalty is moving; the comments tell you why. With a few dozen answers you can read them all. With thousands you need a system.

1. Tag every comment by theme: delivery, price, product quality, app, staff, billing and so on.
2. Score sentiment by theme. One comment often mentions several topics with different feelings: "Love the product, but delivery took two weeks." Our sentiment analysis guide explains how this works.
3. Compare promoters and detractors. What do promoters praise that detractors complain about? Those are your key drivers.
4. Break it down by segment. A theme that drives detractors in one region or channel may be invisible in the overall numbers.
5. Find the root cause. "Delivery was late" is a symptom. Is it one carrier, one warehouse, one region? See our guide to root cause analysis in customer feedback.
6. Track themes over time and alert on sudden changes, not just on the quarterly score.

How to improve your NPS

• Fix the top detractor themes first. Removing reasons for detraction usually moves the score faster than adding delight.
• Follow up with detractors personally and quickly. A resolved problem can turn a detractor into a promoter.
• Protect what promoters love. Their comments show your strengths; don't cut them in the next cost-saving round.
• Work on the weakest segment, not the average.
• Close the loop publicly: "You told us X, we changed Y."
• Connect NPS to operational data (delivery times, ticket resolution, product usage) so you can see which process is behind each theme.

For the wider picture beyond one metric, see what voice of customer analytics is.

NPS vs CSAT vs CES: which should you use?

They are not competitors. A common setup is NPS for the relationship, CSAT after key interactions and CES on high-effort processes, each with an open-ended "why" question.

Limitations of NPS

• One number, many causes. The score does not explain itself; without comments and segments it is hard to act on.
• The average hides segments, as the worked example above shows.
• Cultural and industry differences make cross-company comparisons unreliable.
• Response bias. The customers who answer are not always representative, and frequent surveys reduce response rates.
• Intent is not behaviour. "Likely to recommend" is a stated intention; check it against retention and referral data where you can.

None of these make NPS useless. They mean NPS works best as a starting point, read together with what customers say and do.

Where Pivony fits

• NPS analysis: Pivony breaks NPS down by customer segment, sales channel and region, tracks the trend, shows what drives promoters and detractors, and alerts you when a segment shifts abnormally. It links each movement to the root cause, using NPS comments together with support tickets, app reviews and other feedback.
• Sonic Survey lets you add an open-ended follow-up to your NPS question: branded surveys you embed on your site, with AI analysis of every answer, and a free tier of 25 responses a month.

Next step

Want to see what is behind your NPS? Request a free 24h root cause scan on a sample of your NPS comments, reviews or tickets, or book a demo with your own NPS data.