NPS Report Template with a Worked Example

Identify the exact recommendation question, the product or relationship it refers to, the audience and the collection window. Record when people were invited: asking all active customers on a fixed date produces a different respondent group from asking only people who have just completed onboarding. Retain changes in invitation channel or eligibility alongside the result.

Under the standard 0–10 classification, ratings 9–10 are promoters, 7–8 passives and 0–6 detractors. The score subtracts the detractor percentage from the promoter percentage, using all valid ratings as the denominator. Bain documents this scoring convention in the source below. Keep the counts so reviewers can reproduce the result.

Separate valid ratings from invitations, unanswered questions and unusable responses. An unanswered question is not a passive rating. Document how duplicate responses were handled and whether the unit is a response, a person or an account. Ten responses do not establish ten distinct companies, and a large account may contribute more than one respondent.

Source: Bain & Company: Measuring your Net Promoter Score

Lead with a checkable result, then explain its scope. Report NPS as a score, for example +30, rather than calling it 30% satisfied. Include category percentages as well as counts when useful. State rounding clearly and calculate from the original counts rather than subtracting already rounded displayed percentages.

Treat the interpretation field as a boundary on the claim. 'Among respondents in this window' is supportable from the collected ratings; 'our customers became more loyal because of the release' needs additional evidence. A numerical increase can coexist with a different invitation list, a smaller sample or a changed mix of customers.

Keep survey records and private comments in the appropriate restricted system. A shared internal report can link to that evidence without copying identifiers into every update. Add a named next check so the report leads to an investigation, while keeping an observed theme separate from a promised feature.

This fictional example concerns an imaginary reporting product and is not Feedbackly customer data. In October, a survey receives 100 valid ratings: 55 promoters, 20 passives and 25 detractors. NPS is (55 − 25) ÷ 100 × 100 = +30. The distribution matters: the report contains both a sizeable promoter group and 25 detractor ratings.

The fictional September report had 40 promoters, 30 passives and 30 detractors out of 100, producing +10. October is 20 NPS points higher. Equal response counts alone do not make the comparison valid: check the question, eligible audience, invitation timing and collection route. If October added recently onboarded customers to the list, disclose that difference before explaining the change.

In the fictional comment review, eight of the 25 detractors mention repeated filter setup. That is eight comments within this respondent group, not evidence that 32% of all customers have the problem. The owner will examine the task and existing behavior. The ratings do not establish that a saved-filter feature is the right response, or that another recent release caused the higher score.

For a period comparison, retain both distributions and describe the arithmetic difference in NPS points. Moving from −10 to +10 is an increase of 20 points. A percentage-growth description is confusing across zero and unnecessary for this score. A change in who answered can affect the number even when no individual changed their view.

For a combined result, pool compatible raw counts instead of taking a simple average of group scores. Fictional group A has 8 promoters, 1 passive and 1 detractor: +70 from ten ratings. Group B has 36 promoters, 18 passives and 36 detractors: 0 from 90 ratings. Combined counts are 44, 19 and 37, so the result is +7 from 100 ratings, not the +35 produced by averaging the two scores.

Pool only records that belong to the same defined question and reporting population, with a consistent duplicate policy. Keep subgroup results available: the overall number can hide a change in the balance of larger and smaller groups. Do not mix incompatible questions simply because both use a 0–10 scale.

Before publishing an explanation, review source comments and check plausible alternatives. This template does not calculate uncertainty intervals, test statistical significance or establish that a particular intervention worked. Record a next check that could support or challenge the explanation instead of converting a score movement directly into a delivery commitment.

Take the concrete problems from the comments into your normal feedback analysis. Retain the customer's task, the obstacle and the supporting source. A recommendation rating is context for a conversation, not a priority score for every feature suggested by that respondent.

Feedbackly can hold suitable public-safe product requests, comments, votes and visible progress. Keep survey collection, individual ratings and confidential research in your existing systems. Copying this report does not create a survey, connect a response feed or send a customer message. Assign the investigation and reporting checkpoint in your internal workspace, then communicate actual product decisions through the channel you own.