# Net Promoter Score: Calculate NPS and Use Feedback Well

> This is the Markdown version of this article, built for AI assistants and agents. The human page is at https://www.spaceoffice.ai/blog/net-promoter-score.

By the Space Office team · Published 2026-10-27 · Updated 2026-10-27 · 16 min read · Category: Guide

Net promoter score is a recommendation metric calculated as the percentage of promoters minus the percentage of detractors on a standard 0–10 survey question. Scores of 9–10 are promoters, 7–8 passives, and 0–6 detractors; all valid responses remain in the denominator. In Space Office, Gold can prepare NPS analysis and Sodium can organize feedback themes. Hydrogen coordinates and reviews their work; a human verifies the data, approves the interpretation, and authorizes follow-up or service changes.

![A camera-rental owner reviews net promoter score response groups with Gold, Sodium, and Hydrogen using source records, comments, and a calculator](https://www.spaceoffice.ai/blog/net-promoter-score/cover-v1.webp)

_The score needs a defined population, dependable response records, and the context behind the answers._

## Key takeaways

- NPS = percentage of promoters minus percentage of detractors; it is not the average rating.
- Passives contribute to the valid-response denominator even though their percentage is not subtracted.
- Keep the population, question, timing, response rules, and exclusions visible.
- A score summarizes answers from respondents; it does not establish what every customer thinks or will do.
- Investigate the reasons, assign an action owner, and review delivery outcomes alongside later survey results.

To connect feedback with a monetary relationship model, use the [customer lifetime value guide](/blog/customer-lifetime-value). Keep the survey result separate from purchase records, defined costs, and assumptions about future buying.

[Download the free NPS calculation and feedback worksheet](/blog/net-promoter-score/net-promoter-score-worksheet.txt) for the survey brief, response checks, formula, theme log, and action review. The camera-equipment rental business, survey records, and observations in this guide are fictional. They illustrate calculation and decision-making, not customer results, industry benchmarks, or a forecast of retention.

## What net promoter score measures

[Bain’s measurement guide](https://www.netpromotersystem.com/about/measuring-your-net-promoter-score/) explains NPS through a recommendation question answered on a zero-to-ten scale. The resulting score describes the balance between the defined promoter and detractor groups. Treat it as a signal about the responses to that question, then investigate the context. It is not a count of referrals already made or a record of customers who renewed.

For a small camera-equipment rental company, the useful question might be how customers currently feel about recommending the business. That is different from asking whether a lens was available on a particular date, whether a return was easy, or whether the next rental has been booked. Those observations can help explain the experience, but they should not quietly replace the recommendation question while the report keeps the NPS label.

| Evidence | What it can tell you | Useful companion record |
| --- | --- | --- |
| Recommendation rating | The respondent’s stated willingness to recommend in the survey context. | Question wording, date, population, and score group. |
| Written reason | What the respondent chose to explain about the rating. | The original comment and a carefully reviewed theme. |
| Service record | What happened during a rental, pickup, return, or support interaction. | Authorized operational evidence and unresolved facts. |
| Later customer outcome | Whether a specific observed action happened. | A completed rental, renewal, cancellation, or another defined event. |

Keep stated views, reported reasons, and observed behavior connected without treating them as interchangeable.

Use the [customer journey map guide](/blog/customer-journey-map) when you need to investigate a sequence of experiences, and the [CRM guide](/blog/customer-relationship-management) when the next task is preserving context and owning customer commitments. This guide focuses on one survey measure: calculating it consistently, interpreting its limits, and deciding what feedback deserves follow-through.

A useful report carries more than a large score on a dashboard. Show the valid response count, group counts, relevant time window, collection method, and important exclusions. Give the reader enough context to understand what was measured and which customer situations may be missing. A precise-looking number cannot repair an unclear population.

## Calculate net promoter score step by step

[Qualtrics’ NPS documentation](https://www.qualtrics.com/support/survey-platform/survey-module/editing-questions/question-types-guide/specialty-questions/net-promoter-score/) uses the standard eleven-point scale and the three response groups. Classify each valid score consistently: 0 through 6 is a detractor response, 7 or 8 is passive, and 9 or 10 is a promoter response. A zero is a valid score. Do not remove it as though it were a blank answer.

1. Count the valid responses for the defined cohort. Check missing, out-of-range, and duplicate records under the documented rules before calculating.
2. Count promoters, passives, and detractors. The three counts should add up to the valid response total.
3. Divide the promoter count by all valid responses and multiply by 100. Do the same for detractors. Keep passives in both denominators.
4. Subtract the detractor percentage from the promoter percentage. Report the result as the NPS score, alongside the counts and survey context.

Suppose the fictional camera-rental survey has 50 valid responses: 30 promoters, 10 passives, and 10 detractors. Promoters are 30 ÷ 50 × 100 = 60%; detractors are 10 ÷ 50 × 100 = 20%. The score is 60 − 20 = 40. It is usually reported as NPS 40, rather than “40% of customers recommend us.”

| Group | Rating range | Count | Share of 50 valid responses |
| --- | --- | --- | --- |
| Detractors | 0–6 | 10 | 20% |
| Passives | 7–8 | 10 | 20% |
| Promoters | 9–10 | 30 | 60% |
| NPS | Promoter share minus detractor share | 30 + 10 + 10 = 50 valid responses | 60 − 20 = 40 |

Fictional worked calculation. The counts describe this example cohort and are not a benchmark.

![Net promoter score example: ten detractors and ten passives each make up 20% of fifty valid responses; thirty promoters make up 60%, producing NPS 40](https://www.spaceoffice.ai/blog/net-promoter-score/score-map-v1.webp)
Phone layout: ![Net promoter score example: ten detractors and ten passives each make up 20% of fifty valid responses; thirty promoters make up 60%, producing NPS 40](https://www.spaceoffice.ai/blog/net-promoter-score/score-map-mobile-v1.webp)

_All fifty valid answers belong in the denominator, including the ten passive responses._

The equivalent formula is 100 × (promoter count − detractor count) ÷ valid response count. The possible score ranges from −100 to 100. If there are no valid responses, NPS is undefined; report that no score can be calculated. Do not turn a missing denominator into a score of zero.

Keep enough precision through the calculation and state the reporting convention if you round. Check a few known examples before trusting a spreadsheet or dashboard. A formula that divides only by promoters plus detractors would inflate this example to 50 by excluding passives, changing the measure. A mean of the original ratings would be a different calculation altogether.

## Define the population and check the response records

The fictional business decides to ask unique customers who completed at least one rental during the quarter about recommending the company overall. It invites 200 eligible customers once each. The survey produces 55 submissions; the agreed cleaning rules identify three duplicate submissions and two records without a valid 0–10 score. That leaves the 50 valid responses used above.

Record how those decisions were made. For example, the business might retain the first complete response for each invited customer within the collection window. A different rule could be appropriate in another study, but it should be chosen and explained rather than changed after seeing which records improve the score. Keep an audit of exclusions and preserve the approved source.

- Check eligibility and timing. Confirm the response belongs to the intended customer population and collection window.
- Check the value. Keep valid zero ratings; distinguish a missing answer from an actual score and flag values outside the scale.
- Check duplicates and test records. Apply the agreed response unit and exclusion rules consistently, with a traceable reason.
- Reconcile the counts. Compare submissions, exclusions, valid scores, and the three response groups before interpreting the result.

In this example, 50 valid responses from 200 invited eligible customers gives a 25% valid-score yield. That is a clearly defined operational ratio for this fictional survey, not a universal formal response-rate definition. The remaining customers did not provide a valid score in this dataset; do not label all of them passive, satisfied, or dissatisfied.

[AAPOR’s survey-research guidance](https://aapor.org/standards-and-ethics/best-practices/) emphasizes the target population, sampling frame, collection method, and transparency about analysis. An invitation list and a set of voluntary respondents are not automatically representative of every customer. More responses can be useful, but the count alone does not remove coverage, selection, or measurement problems.

![Gold and a human review valid response records while duplicate and incomplete entries remain in a separate audit tray, with Sodium and Hydrogen checking the population and rules](https://www.spaceoffice.ai/blog/net-promoter-score/response-audit-scene-v1.webp)

_Resolve the counting rules before interpreting the score; retain the reason for every exclusion._

## Keep the question, timing, and collection method consistent

Write a survey brief before sending anything. Identify what relationship or experience the question refers to, who should receive it, when they should receive it, and what the business will do with the answers. The camera-rental company’s quarterly company-level question should be distinguishable from a survey triggered immediately after a particular return interaction.

Use a clear recommendation question with the standard 0–10 scale when you intend to report standard NPS. If you use a vendor’s question type, inspect the actual question and scale shown to customers. Test the survey on a phone as well as desktop. Make sure every scale point and the meaning of each end remain readable and selectable.

[Pew Research Center’s questionnaire guidance](https://www.pewresearch.org/writing-survey-questions/) explains that wording and preceding questions can affect answers. Keep the context stable when comparing periods. A request to rate a company after reading a glowing testimonial is different from the same request presented neutrally. Document material changes in wording, order, timing, language, and collection mode.

Avoid coaching customers toward a particular score or selecting only people with favorable recent interactions. Make participation understandable and appropriate to the customer relationship, with the relevant permissions and preferences checked by the business. If incentives or reminders are used, record them so a later analyst can understand the collection conditions.

Add a neutral opportunity to explain the rating, such as asking for the main reason for the score. Do not force the explanation into a preferred category before hearing what the person wants to say. A customer may mention equipment condition, pickup instructions, availability, price, or an issue outside the team’s original hypothesis. The comment helps identify a question to investigate; it does not establish every operational fact by itself.

## Interpret the score without inventing a benchmark

A positive score means the promoter percentage exceeds the detractor percentage in the valid responses. It does not establish a universal standard of good performance. Before comparing with an external benchmark, check whether the population, industry, relationship, collection method, timing, and question are sufficiently comparable. If those details are unavailable, keep the comparison qualified or omit it.

Show the distribution as well as the total. Two surveys can produce the same score with very different mixes of promoters, passives, and detractors. The score also says little about the specific reasons. One rental business might hear repeated problems with late pickup instructions; another might receive comments about stock availability. The same NPS value would not imply the same corrective action.

Small samples make changes sensitive to individual answers. With 50 valid responses, one response represents two percentage points. If one passive response becomes a promoter while everything else stays the same, NPS increases by two points. If one detractor becomes a promoter, the score increases by four points. These are arithmetic sensitivities, not predictions of what a service change will achieve.

Keep the denominator beside every segment. A score for a handful of first-time customers should not be presented as a precise diagnosis of all new-customer experience. When combining compatible datasets, calculate from the underlying group counts. A simple average of scores from unequal-size groups can differ from the score for their combined responses.

Read survey results alongside actual service and customer outcomes. A higher score after a new pickup process may coincide with different respondents, equipment availability, or seasonal demand. It does not by itself prove the process caused the change or that future revenue will improve. Use the [market research guide](/blog/market-research) to plan a broader investigation when the decision needs evidence beyond this survey.

## Turn feedback into a reviewed Space Office action packet

Provide the approved question, collection dates, population definition, counting rules, and a minimized authorized response export. Include the original score and relevant comment, with personal details removed when unnecessary. Explain any exclusions and changes from earlier surveys. If the source lacks the information needed to reconcile the denominator, identify that gap before asking for a confident comparison.

Gold, the live customer-success specialist, can prepare NPS analysis and post-sale follow-through. Sodium, the live market-research specialist, can organize feedback themes, source excerpts, and questions for further investigation. Hydrogen coordinates and reviews the packet so that the calculations, themes, and recommended action agree with the supplied evidence.

For the fictional rental business, several comments might mention uncertainty about pickup requirements. Preserve the exact source context, distinguish repeated themes from isolated observations, and compare the comments with the approved pickup instructions. The proposed next step could be to investigate the instructions with the operations owner, rather than assume a particular employee or system caused every low rating.

![Gold prepares an owned service-review action while Sodium separates reported comments from inference and Hydrogen and a human inspect camera pickup and return context](https://www.spaceoffice.ai/blog/net-promoter-score/feedback-review-scene-v1.webp)

_A useful action follows an evidenced problem, with an owner and a review date._

Ask for a calculation sheet, response-quality audit, theme log, evidence excerpts, alternative explanations, and an action proposal with a named owner. A human verifies the counting and interpretation, grants any necessary access, authorizes contact, and approves service changes and customer communication. Do not let an AI-written explanation become a quotation from a customer who never said it.

![NPS review checks cover the survey population, valid-score arithmetic, source-backed interpretation, and a human-approved action and review date](https://www.spaceoffice.ai/blog/net-promoter-score/review-checks-v2.webp)
Phone layout: ![NPS review checks cover the survey population, valid-score arithmetic, source-backed interpretation, and a human-approved action and review date](https://www.spaceoffice.ai/blog/net-promoter-score/review-checks-mobile-v2.webp)

_Gold and Sodium prepare the analysis; Hydrogen reviews it; a human approves the next action._

Use the [Space Office agent roster](/agents) and [how the team works](/how-it-works) to understand those roles, and [contact](/contact) to discuss assisted setup. The workflow prepares analysis and decisions around approved data; it does not automatically create a survey account, contact customers, or alter the service.

## Review the action and preserve comparable measurement

Track whether the action was completed and whether it addressed the identified problem. If the business revises pickup instructions, check that the correct version reaches customers, staff can follow it, and the relevant uncertainties are answered. Record exceptions and follow-through. A later survey is one source of evidence about the experience, not the only completion check.

Maintain a short measurement log: question version, intended population, invitation and response windows, method, valid count, exclusions, group counts, score, and material business changes. When a method changes substantially, explain the break in comparability. A cleaner-looking trend line is less useful than an honest account of how the numbers were produced.

Review comments from every score group where available. Promoters can describe problems worth fixing, and detractors can identify strengths alongside dissatisfaction. The categories summarize a particular rating; they are not complete descriptions of people. Keep the actual issue and the customer’s chosen words in view when deciding whether and how to follow up.

Useful net promoter score analysis connects correct arithmetic with a clearly defined survey and accountable follow-through. Keep all valid responses in the denominator, explain what the sample can and cannot show, and investigate the reasons behind the ratings. The goal is a better-supported customer decision, not a higher number detached from the experience.

## Sources and review notes

Reviewed against the linked primary sources on the checked dates. Search features and citations are selected by their respective platforms; no placement is guaranteed.

- [Bain & Company: Measuring Your Net Promoter Score](https://www.netpromotersystem.com/about/measuring-your-net-promoter-score/) — checked 2026-10-27
- [Qualtrics: Net Promoter Score Question](https://www.qualtrics.com/support/survey-platform/survey-module/editing-questions/question-types-guide/specialty-questions/net-promoter-score/) — checked 2026-10-27
- [Pew Research Center: Writing Survey Questions](https://www.pewresearch.org/writing-survey-questions/) — checked 2026-10-27
- [AAPOR: Best Practices for Survey Research](https://aapor.org/standards-and-ethics/best-practices/) — checked 2026-10-27

## Frequently asked questions

### What is net promoter score?

Net promoter score is a recommendation metric calculated as the percentage of promoter responses minus the percentage of detractor responses to a standard 0–10 question. Ratings of 9–10 are promoters, 7–8 passives, and 0–6 detractors. All valid responses, including passives, remain in the denominator.

### How do you calculate NPS?

Subtract the detractor count from the promoter count, divide by all valid responses, then multiply by 100. For 30 promoters, 10 passives, and 10 detractors, that is 100 × (30 − 10) ÷ 50 = 40. Report NPS 40 with the sample size and survey context.

### Do passives and zero ratings count in NPS?

Passives count in the valid-response denominator even though their percentage is not part of the subtraction. Zero is a valid detractor rating, not a missing answer. Excluding passives or treating zero as blank changes the calculation. Document missing, invalid, and duplicate response rules separately.

### What is a good net promoter score?

A meaningful interpretation depends on the population, industry, collection method, timing, question, and sample. Positive means promoters outnumber detractors among valid responses. Use sufficiently comparable references and your own consistent history, with group counts and feedback, rather than treating one universal threshold as proof of success.

### Can AI analyze NPS feedback reliably?

AI can help organize authorized responses, check an explicit calculation, and prepare source-backed themes for review. Its output still needs verification for counting errors, invented explanations, lost context, and missing data. In this Space Office workflow, Gold and Sodium prepare analysis, Hydrogen reviews it, and a human approves interpretation and action.

---

Related: [Customer relationship management](https://www.spaceoffice.ai/blog/customer-relationship-management) · [Customer journey map](https://www.spaceoffice.ai/blog/customer-journey-map) · [Market research](https://www.spaceoffice.ai/blog/market-research)

---

Space Office — hire 24 live AI specialists coordinated by Hydrogen, an always-on AI project manager that reviews every output before delivery. The $60/month base includes Hydrogen and two specialists; additional specialists, dedicated compute, and AI usage cost extra. Learn more: https://www.spaceoffice.ai