Net Promoter Score (NPS): how to measure and improve customer loyalty

Net Promoter Score (NPS) measures customer loyalty with one question: how likely are you to recommend us, on a scale of 0 to 10. Subtract the percentage of detractors (0 to 6) from the percentage of promoters (9 to 10) and you get a score between -100 and +100.

Anything above 0 means you have more promoters than detractors, while a score above 50 is generally considered excellent. But the score only tells you part of the story. The open-ended follow-up is where you learn what to actually fix.

Net Promoter Score is a customer loyalty metric that measures how likely your customers are to recommend your product or service to someone else, on a scale of 0 to 10.

Fred Reichheld, a partner at Bain & Company, introduced NPS in a 2003 Harvard Business Review article called “The One Number You Need to Grow.” The goal was to create a simpler alternative to lengthy customer satisfaction surveys and a metric teams could use to track customer loyalty over time. Two decades later, NPS is widely used across industries, from SaaS to retail to healthcare.

The appeal is that it is simple to run and simple to read. One question gives you a number you can track over time and compare with relevant benchmarks. The catch, which we will come back to, is that the number on its own does not tell you what to do next.

The standard NPS question is: how likely is it that you would recommend this company or product to a friend or colleague?

Respondents answer on a 0 to 10 scale. That single rating gives you the data you need to calculate your NPS.

But the rating is only half of a good NPS survey. The second question, the open-ended one, is where you learn why someone chose that score: what is the main reason for your score?

A score of 4 tells you a customer is unhappy, but not why. “A 4, because your onboarding took three weeks and no one followed up” tells you where to investigate and what may need to change. Skip the follow-up and you are left with a trendline and no explanation for it.

Keep the survey to these two questions when they are all you need. Part of NPS's appeal is its simplicity, and adding too many questions can create unnecessary friction. You can grab a copy of our NPS survey template to start from a proven structure.

The formula is: NPS = % promoters minus % detractors.

Passives are counted in your total responses but left out of the subtraction. Here is how it works with a small sample. Say you collect 100 responses:

The 0 to 10 scale sorts every respondent into one of three groups.

There are two ways to judge your score.

The absolute method compares your number against a general standard. As a simple rule of thumb:

Benchmarks shift year to year, so treat these as directional rather than fixed targets. These figures come from Retently's 2026 NPS benchmark study.

There are two ways to run NPS, and mixing them up muddies your data.

Relational NPS asks about the overall relationship on a fixed schedule, such as quarterly or twice a year. It is a health check on the whole customer relationship and the version most people mean when they say “our NPS.”

Transactional NPS is collected after a specific interaction, such as a support ticket, onboarding call, or purchase. Based on your support experience today, how likely are you to recommend us? It helps you understand how customers perceive a particular touchpoint.

Use relational NPS as your headline loyalty metric and transactional NPS to understand specific moments in the customer journey. Keep the results separate, since they measure different things and should not be combined into a single NPS.

For many SaaS companies, quarterly relational NPS is a useful starting cadence. It is frequent enough to track changes over time without asking customers the same question every few weeks.

But there is no universal NPS schedule. The right cadence depends on your customer lifecycle, how often people use the product, and how quickly their experience is likely to change. Semi-annual surveys may make more sense for lower-touch or slower-moving customer relationships, while higher-frequency products may benefit from more regular measurement.

Transactional NPS runs on a different clock: send it shortly after the interaction it is measuring, while the experience is fresh.

Calculating the score takes a minute. Making sense of it is the real work, and much of that happens in the open-ended responses.

Start with the number so you know where you stand, then read the verbatims by segment. Group them by score band first: what do your detractors keep saying, and what do your promoters credit? Patterns matter more than any single comment. If ten detractors all mention the same missing feature, that is a signal worth investigating, especially if it lines up with what you are hearing elsewhere in your research. Our customer feedback analysis guide covers how to code those verbatims consistently.

This is where NPS analysis usually gets harder at scale. A few dozen verbatims are readable by hand. A few thousand are not, so teams need a reliable way to organize and synthesize the responses without losing the context behind them.

That is the gap Great Question is built to close. Run your NPS survey in Great Question, then use AI theme clustering to group open-ended responses automatically. Recurring themes like billing confusion or slow onboarding surface as patterns instead of disappearing into a wall of text. Because the results sit alongside your interviews and other studies in a research repository, you can also connect what you hear in NPS responses with the rest of your customer research.

Dedicated NPS tools are good at firing the survey and charting the trendline. But if the score lives in one system and the rest of your customer research lives somewhere else, understanding the why can still take manual work. Keeping the score and the reasons behind it in one research platform makes it easier to move from measurement to investigation and action. For a wider view of options, see our roundup of product feedback tools.

You do not improve NPS by chasing the number. You improve it by understanding what is driving the scores and acting on the problems you find.

NPS is useful, but it is not the whole picture, and understanding its limitations makes the metric more useful.

It is a single number, so it flattens a lot of nuance. Two companies with an NPS of 40 can have completely different customer bases and problems. Stated intent to recommend is also not the same as actually recommending, and the two do not always line up. Sample size and who responds matter too. A small or unrepresentative sample can produce a score that looks more definitive than it really is.

None of this makes NPS worthless. It makes it a starting point. Use it as a fast, trackable pulse on loyalty, then pair it with the qualitative why and with harder behavioral data like renewals and referrals. NPS can tell you that customer sentiment is changing. The follow-up responses and other research help you understand why. Our guide to analyzing survey data goes deeper on reading results responsibly.

Employee Net Promoter Score applies the same 0 to 10 framework to your own team: how likely are you to recommend this company as a place to work?

The math is identical. The interpretation requires its own context. Employee experience is more complicated than a single recommendation score, and concerns about anonymity or how feedback will be used can influence how candidly people respond. Treat eNPS as one signal alongside other employee feedback rather than a standalone measure of engagement or satisfaction. If you use it, the open-ended follow-up is essential for understanding what is driving the score.

NPS is a fast way to take the pulse of customer loyalty, but the score is only the starting point. Understanding why customers gave that score is what makes the feedback actionable.

Grab our free NPS survey template to launch your first survey, and read our survey research guide for how NPS fits alongside the rest of your customer research.