How to Calculate NPS: 0–10 Categories, Examples, and Interpretation
Published:
Last Updated:
Category: Marketing Glossary, CRM, LTV & Customer Management
Authors: Shusaku Yosa
NPS, or Net Promoter Score, measures stated willingness to recommend. Classify answers on a 0–10 scale, then subtract the percentage of detractors from the percentage of promoters. The equivalent count formula is (promoters − detractors) ÷ valid responses × 100. Passives remain in the denominator.
The result is a score between −100 and +100, not a percentage of customers. Correct arithmetic does not make respondents representative of every customer. This guide covers calculation and interpretation; use the NPS survey design guide to define the audience, questions, collection, and follow-up.
Classify the eleven response options
An NPS question asks how likely someone is to recommend the product or service to a friend or colleague, using eleven options from 0 through 10. Bain defines the standard categories as promoters at 9–10, passives at 7–8, and detractors at 0–6.
Score | Category | Treatment in the difference | Valid-response denominator |
|---|---|---|---|
9–10 | Promoter | Add the count | Included |
7–8 | Passive | Neither added nor subtracted | Included |
0–6 | Detractor | Subtract the count | Included |
Missing or out of range | Review separately | Not used | Apply the pre-agreed validity rules; record exclusions |
Passives are absent from the numerator's subtraction, but not from the total response count. A 1–10 scale or a five-point satisfaction scale cannot simply be treated as the standard NPS scale.
Calculate a fictional example with 100 responses
Suppose a fictional survey has 100 valid responses: 40 promoters, 35 passives, and 25 detractors.
Measure | Count or share | Calculation |
|---|---|---|
Promoters | 40; 40% | 40 ÷ 100 × 100 |
Passives | 35; 35% | 35 ÷ 100 × 100 |
Detractors | 25; 25% | 25 ÷ 100 × 100 |
NPS | +15 | (40 − 25) ÷ 100 × 100 |
The difference between 40% and 25% produces an NPS of +15. It does not mean 15% of customers recommend the service: the promoter share in this example is 40%. Excluding passives from the denominator would produce (40 − 25) ÷ 65 × 100, or about +23.1, which is incorrect.
The average of the original 0–10 ratings is also a different measure. Two surveys can have the same average rating but different distributions across the NPS categories.
Handle zero responses and invalid records explicitly
If there are no valid responses, NPS is undefined, not zero. Report “not measured; zero valid responses.” A genuine score of zero would indicate equal promoter and detractor shares. An answer of 0 is a valid detractor response and must not be mistaken for a missing answer.
Before aggregation, decide how to handle duplicate responses, ineligible respondents, missing ratings, and out-of-range values. Record exclusion counts and reasons. Do not remove low scores merely because they are inconvenient. If a survey design requires weighting, explain the weights and calculation separately from this unweighted example.
Express changes in points and pool using counts
If a comparable later survey produces +25 after +15, the change is +10 points. It is not a ten-percent increase in satisfaction. Report the response counts and collection conditions alongside both results.
When combining comparable groups, recalculate from their category counts rather than averaging their displayed scores. The following fictional groups assume compatible questions, populations, and collection timing.
Group | Valid responses | Promoters / passives / detractors | NPS |
|---|---|---|---|
A | 100 | 40 / 35 / 25 | +15 |
B | 50 | 10 / 20 / 20 | −20 |
Combined | 150 | 50 / 55 / 45 | (50 − 45) ÷ 150 × 100 ≈ +3.3 |
The simple average (+15 − 20) ÷ 2 = −2.5 ignores the different group sizes. The pooled result is approximately +3.3. Keep the subgroup results visible as well: a combined figure can hide a problem concentrated in one customer group.
Check conditions before judging a score
A universal “good score” is not useful without a source, industry, region, question, and collection method. Begin with your own comparable series. For external comparisons, check who was asked, their experience, the wording, and when and how answers were collected.
Bain distinguishes types of NPS, including relationship and experience measures. A survey immediately after onboarding is not directly interchangeable with one about an established relationship. Record changes in language, invitation channel, eligibility, and timing rather than presenting a changed method as a seamless trend.
Respondents may differ from nonrespondents. People with strong opinions might be more likely to answer, while former customers may be unreachable. As AAPOR explains, a response rate alone does not establish the presence or absence of nonresponse bias. Report the eligible audience, invitations, valid responses, and relevant customer characteristics. Small subgroup counts deserve particular caution.
Turn the result into a specific investigation
- Store the survey period, eligibility rules, question, collection method, and valid count.
- Reconcile the three category counts to all valid responses.
- Review reasons and changes in respondent composition, not just the score.
- Avoid excessive segmentation that creates tiny or identifiable groups.
- Measure recommendation intent, actual referrals, retention, purchases, and gross profit separately.
- Do not attribute a before-and-after score change solely to one intervention.
NPS does not replace revenue or retention measurement. Connect the investigation to customer lifetime value initiatives or OMO experience design where relevant. To organize improvement initiatives, owners, and evaluation measures, explore Xtrategy's features.
