Back to the subscription dictionary

Net Promoter Score (NPS)

Net Promoter Score (NPS) summarises respondents’ stated likelihood of recommending a product or company on a 0–10 scale. Subtract the share answering 0–6 from the share answering 9–10. The result ranges from −100 to +100 points and describes the responses, rather than certain future behaviour.

Prepared with AIContent updated

One recommendation question, three response categories

Respondents rate their likelihood of recommending the chosen company or service to a friend or colleague on a 0–10 scale. Keep the subject, wording and scale consistent when comparing surveys. A question after a support interaction need not measure the same thing as a general assessment of the service.

Responses of 9–10 are called Promoters, 7–8 Passives and 0–6 Detractors. These are categories of stated responses. A category does not establish that a particular person will refer others, renew, spend more or publish negative comments.

Two hundred responses, three groups

Illustrative survey with all 200 valid responses in the denominator.

NPS = 120/200 × 100 − 30/200 × 100 = 60 − 15

Promoters: 9–10
120 responses
Passives: 7–8
50 responses
Detractors: 0–6
30 responses
NPS45 points

NPS is a score from −100 to +100, not a percentage. Passive responses remain in the denominator although their share is neither added nor subtracted.

Calculate using every valid response

Assume 200 valid responses: 120 in the 9–10 group, 50 in 7–8 and 30 in 0–6. Their shares are 60%, 25% and 15%. NPS = 60 − 15 = +45 points.

Passive responses remain in the denominator although their share is not separately added or subtracted. The score does not mean that 45% of customers are satisfied or that a known number will actually recommend the service. A positive score simply means the 9–10 share exceeds the 0–6 share.

Sample and timing affect the result

Document who was invited, how many responded, and the channel, language and timing. A small or unrepresentative group can differ from the wider customer population. Changing the survey timing can also change the mix of respondents.

Choose a frequency suited to the purpose without overburdening respondents. Compare equivalent surveys and read the comments. There is no universal good SaaS score; +45 is an arithmetic example here, not a benchmark.

Use comments to investigate specific issues

Ask why the rating was chosen and identify themes such as functionality, delivery, support or price. A low rating may justify a conversation, but does not by itself mean the person will cancel. Passive and positive responses can also identify useful problems.

Use the findings to formulate and test improvements. Assess changes using comparable groups, later feedback and actual outcomes. A higher score alone does not establish that a particular action caused an improvement.

NPS and subscription economics answer different questions

Observed retention and churn follow defined customer cohorts and contract events. MRR and ARR measure normalised recurring contract value. CLV is a model estimate, while CAC uses a defined allocation of acquisition costs. None can be calculated directly from NPS.

It may be useful to investigate whether response groups subsequently renew at different rates. Document populations and periods and consider nonrespondents. An association does not establish a cause, and a recommendation score is not a revenue forecast.

Report the response base with the score

For example, report NPS +45 with 200 valid responses, the response rate, measurement period and main comment themes. Separate the score, observations and proposed actions. It should be possible to recalculate the result from the three response categories.

Track relevant customer and operational outcomes separately. Use the actual feedback to investigate possible improvements, rather than inferring loyalty or financial returns from category names.

Questions and answers

How is NPS calculated?

Calculate the share of all valid responses scoring 9–10 and subtract the share scoring 0–6. Passive responses of 7–8 remain in the denominator. Shares of 60% and 15% give NPS +45 points, not 45%.

Does a high score prove customer loyalty?

No. It describes stated recommendation likelihood among respondents. Actual referrals, renewals and cancellations require separate observations.

Can NPS calculate MRR, CLV or CAC?

No. Investigate associations in relevant cohorts, but use contract, cost and behaviour data for the financial measures. Correlation does not identify which action caused a financial outcome.

What is a good NPS?

Use a relevant comparison with the same question, audience, timing and method. No threshold fits every SaaS product. Report sample size and response rate with the score.

Can weekly or segment NPS scores be added?

No. Recalculate a combined score from the relevant response counts under a consistent method. Giving equal weight to segments with different response counts need not match the score calculated from all responses together.

Sources and documentation

We update our content regularly and do our best to keep it accurate. If you find an error or think something is missing, please tell us. We are always looking for ways to improve the content.

Found an error? Let us know