Survey: NPS (Net Promoter Score)

EvaluationquantitativeBeginner

TL;DR

The Net Promoter Score (NPS) is a single-question survey that measures the stated likelihood that someone would recommend a company, product or service.

Detailed description

NPS asks a single question: how likely it is that the person would recommend the company, or the product in many applications, to a friend or colleague. It is answered on a scale from 0 to 10, where 10 is extremely likely and 0 is not at all likely. The calculation Reichheld (2003) proposed groups the responses and subtracts two percentages:

Promoters: those who answer 9 or 10.
Passives: those who answer 7 or 8.
Detractors: those who answer 0 through 6.
The score is the percentage of promoters minus the percentage of detractors, and it is expressed as a percentage.

There is no universal threshold for a good NPS: reference figures depend on which products are in the sample. Reichheld (2003) reported a median of 16% across more than 400 companies in 28 industries, with data collected by Satmetrix more than twenty years ago. MeasuringU's measurements of consumer software, with participants almost all from the United States, show how much that average moves: 23% in 2017, 1% in 2020, 4% in 2022 and 24% in 2025, with between 17 and 41 products per wave. Sauro and Lewis (2020) consider it more likely that the drop reflects the increase in the number of products evaluated than a change in loyalty. It is a practitioner source, not peer reviewed.

What does repeat across the four waves is the relationship with perceived usability: people who would recommend score higher on the SUS than people who would not. In 2025 promoters averaged 87.0, passives 73.5 and detractors 61.2, and the same order appears in 2017, 2020 and 2022.

NPS asks about willingness to recommend, which Reichheld (2003) proposed as an indicator of loyalty toward the company. Fisher and Kordupleski (2019) describe how in practice it is often requested right after a transaction, and they criticize that use because a one-off annoyance ends up weighing on the assessment of the whole relationship. Satisfaction is a different construct: it can be measured on a specific interaction or on accumulated experience (ACSI, 2008), and Oliver (1980) models it as the result of prior expectations and of whether the experience exceeded them or not. Reichheld (2003) himself limits where his question works: he reports that it was not the best predictor of growth in database software or in sectors dominated by monopolies and near monopolies, where the consumer has little choice.

Main objective

The goal is to summarize in a single indicator how willing customers say they are to recommend a company or a product, and to track that over time.

Use cases

Ongoing measurement programs of the relationship with the productTracking by customer segment or by marketComparison across periods within the same user baseComparison with an industry study, when a comparable one existsRelationship surveys paired with an open-ended question

When to use it

When the aim is to track over time the stated willingness to recommend a product or a brand.

Effort level

Low

Recommended number of users

Depends on the margin of error sought; more to compare groups

Advantages

A single question: It takes seconds to answer and can be run repeatedly without building a questionnaire.
One indicator: It summarizes the distribution of responses into a single number that can be tracked over time within the same user base.
Simple to communicate: According to Bendle and Bagga (2016), its simplicity is one of its strongest arguments in its favor.
Can be segmented: The same calculation applies by region, channel or customer segment, and those can be compared with each other.
Associated with perceived usability: In MeasuringU's measurements of consumer software, people who would recommend score higher on the SUS than people who would not.

Disadvantages

Not diagnostic: It shows how the relationship is going, not what to fix or in what order.
Debatable cut-offs: According to Bendle and Bagga (2016), the boundaries between 6 and 7, and between 8 and 9, seem somewhat arbitrary and culturally dependent.
Grouping loses information: Someone who answers 0 and someone who answers 6 both count as the same detractor, although they probably do not behave the same way (Bendle and Bagga, 2016).
It can rise without the experience improving: If detractors stop being customers, the score improves even though nothing has changed (Fisher and Kordupleski, 2019).
Superiority not demonstrated: Bendle and Bagga (2016) note that they know of no rigorous studies showing that NPS is superior to other customer experience metrics.

When to use

•To track reported loyalty toward a product or a brand over time
•To compare the same user base across periods, segments or markets
•When a single indicator is needed that leadership can track alongside business metrics

Metrics

•Recommendation scale from 0 to 10
•Percentage of promoters, passives and detractors
•NPS: percentage of promoters minus percentage of detractors
•Score range: from -100% to 100%
•Number of responses in the measurement
•Difference from the previous measurement of the same base

Practical example

Illustrative example: in a measurement with 400 responses, 45% are promoters, 30% passives and 25% detractors. The score is 45 minus 25, that is 20%.

Related methodologies

Related Resources

Tool

Sample Calculator

Calculate how many participants you need for your research

Service

Need to run this study?

Our team designs, executes and analyzes this methodology for your project