What is a good NPS score?
A good NPS score needs a relevant reference: comparable competitors, a well-matched industry study, your own previous results or a defined improvement goal. There is no single cutoff that establishes good performance for every organization and survey.
A positive score means promoters outnumber detractors among valid responses. It does not by itself mean performance is strong for your market. A score can improve and still leave important problems unresolved. Equally, an external benchmark can provide useful context even when it does not explain every difference.
Use this guide to choose a reference, check comparability and turn the comparison into a useful review. The aim is to understand where attention is needed, rather than find a number that makes a report look favorable.
Published industry NPS examples
The following are selected historical examples from the Qualtrics XM Institute Q3–Q4 2024 U.S. Consumer Benchmark Study. The published chart covers 10,000 U.S. consumers, 354 organizations and 22 industries. These are not 2026 targets, global standards or B2B benchmarks.
Scroll horizontally to see all columns →
| Industry in the source | Published NPS |
|---|---|
| Software firm | 21.1 |
| Bank | 28.0 |
| Consumer payment | 31.5 |
| Grocery | 34.3 |
Source: 2024 industry benchmark chart. The provider also lists a 2025 consumer report; its public landing page does not expose the detailed industry values used here. Check the relevant report and methodology before setting a target.
The examples illustrate why the market and study context matter. They should not be used to declare that a small B2B software provider with an account-contact survey is above or below the consumer software market. The audience, collection method and unit of reporting may differ substantially.
Choose the right kind of comparison
Bain distinguishes competitive benchmark, relationship and experience NPS. Competitive research examines relative market standing; relationship feedback concerns the broader customer relationship; experience feedback follows a particular interaction. See Bain’s explanation of the three types.
Scroll horizontally to see all columns →
| Reference | Useful question | Main check |
|---|---|---|
| Comparable competitive study | How do customers view us relative to alternatives? | Are the brands assessed through a consistent research design? |
| Industry report | What wider context is available? | Does the population, geography and survey type fit? |
| Internal trend | How has our measured feedback changed? | Have audience, response patterns and collection conditions changed? |
| Internal segment comparison | Where should we investigate differences? | Are groups defined consistently and large enough to interpret? |
Do not mix an immediate post-support score with an annual relationship score merely because both use a 0–10 scale. Likewise, a self-published company figure with no methodology is a weaker reference than a documented study designed for comparison.
What should you check in an NPS benchmark?
- Audience: consumers, businesses, members, users or account contacts.
- Geography and language: the market and questionnaire versions represented.
- Survey purpose: relationship, interaction or competitive research.
- Collection period: when the experience and responses were measured.
- Sampling and response: who was invited, how they were selected and who answered.
- Wording and channel: the question, scale and delivery method.
- Aggregation: whether figures represent people, accounts, brands or another unit.
- Uncertainty: sample sizes, weighting and any reported intervals or limitations.
If key information is missing, label the comparison as limited context. Do not silently fill gaps with assumptions. A number from a different market may still be interesting, but it should not carry the same weight as a directly relevant study.
Build an internal baseline you can maintain
Record the rating counts and NPS alongside the audience definition, invitation count where meaningful, valid-response count and period. Keep the exact questionnaire version and relevant collection conditions. This makes it possible to investigate later changes.
Stable wording and timing help comparability, but they do not remove every source of variation. Nonresponse, changing customer mix, seasonality and other events can still affect the result. An internal trend is not automatically a controlled experiment.
Where repeat named feedback is appropriate, report matched respondents separately from the full wave. Repeated cross-sectional surveys can also be useful without identifying every person. Choose the design around the research purpose and preserve the promised privacy arrangement.
Worked example: an improving score with a changing audience
In a fictional first wave, 100 customers respond: 50 promoters, 30 passives and 20 detractors. NPS is 30. In a second wave, 100 respond: 55 promoters, 30 passives and 15 detractors. NPS is 40, a ten-point increase.
Before concluding that the experience improved, the team finds that the second wave includes more long-tenured customers and fewer new customers. It should inspect those groups separately and review how invitations and responses changed. The aggregate improvement may reflect experience, audience composition or both.
Suppose only 40 customers answered both waves. Their matched results answer a narrower question than the full-wave scores. Show that group’s counts and coverage rather than treating it as representative of all customers automatically.
If onboarding complaints are less frequent in the second wave, that is useful evidence to investigate. It is not conclusive proof that an onboarding fix caused the score increase, particularly if fewer new customers responded.
Is a small NPS difference meaningful?
The answer depends on the data and design. A few changed responses can move the score substantially in a small sample. For example, with 20 valid responses and a fixed denominator, one person moving from passive to promoter changes NPS by five points.
Do not describe every two-point change as improvement or decline in the underlying customer population. Consider sampling uncertainty, response patterns and practical importance. Where statistical inference is needed, use a method appropriate to the survey design rather than inventing a universal margin of error.
A large sample can still have systematic bias. More responses from the same narrow audience do not necessarily solve coverage problems. Show the limitations that matter for the decision, not only the sample size.
Set a target connected to work the team can do
Use relevant benchmarks and internal evidence together. Identify the experience you want to improve, the group affected and the actions the team can take. A numerical target should sit alongside that plan.
For example, a team might investigate setup delays, assign an owner to the recurring problem and review both operational completion times and customer feedback after the change. NPS supplies one part of that review; it is not a substitute for checking the actual issue.
Avoid incentives that encourage staff to ask only happy customers or pressure people for high ratings. Monitor the invitation rules and who is missing from the data. Comment analysis alone cannot reliably reveal every instance of selective collection.
Keep comparisons consistent across teams
Agree on a small shared core for the comparisons you need, such as the question version, date, survey purpose and selected customer segments. Local teams can ask additional questions relevant to their work. Document which measures can be combined and which should remain separate.
Keep a data dictionary for segment membership and calculations. If a customer changes plan or region, retain the context appropriate to the response date. Otherwise an updated profile can accidentally rewrite the apparent historical segment.
For B2B programs, decide whether the reported figure weights each response or uses a defined account-level approach. Several contacts at a large account should not silently change the meaning of a comparison with a single-contact account survey.
What should an NPS comparison report contain?
Show the current score, the reference score and the point difference. Then show why the reference was selected, its date and the relevant methodological differences. Add response counts, coverage and uncertainty where available.
Present reviewed comment themes as evidence about respondents’ explanations. Include contrasting or unresolved findings where they matter. Do not label a theme as the proven cause of the gap solely because it appears frequently.
A useful conclusion might read: “Our relationship score increased ten points, but the respondent mix changed. We will review new-customer feedback separately and test whether setup delays declined.” That is more actionable than an unsupported declaration of “world-class NPS.”
Connect the benchmark to your feedback workflow
Sopact’s collection, analysis and governance approach is relevant when teams need to keep recurring feedback, definitions and source context available for review. Test whether your proposed workflow can reproduce a score, explain a segment and preserve a change in the questionnaire or analysis rules.
For the buying requirements, use NPS survey software. For collection design, see NPS survey questions. For responding to low ratings, continue to NPS detractors.
Watch: review the feedback behind the score
This video discusses reviewing incoming feedback and its context. Use it alongside the comparison checks above; source-linked comments help interpretation but do not establish benchmark comparability by themselves.
Frequently asked questions
Is an NPS above zero good?
It means promoters outnumber detractors among valid responses. Whether that is strong performance depends on a relevant reference, the survey design and the decision being made.
Should we use an external benchmark?
Yes, when it is relevant and its methodology is sufficiently clear. Use it with your internal evidence and explain material differences in audience, period and survey type.
Is our own trend always reliable?
No. Stable methods help, but changing response patterns, customer mix and other influences can still affect results. Inspect the data behind the trend.
Can relationship and experience NPS be compared directly?
They concern different scopes and collection conditions. Keep them separate unless a justified analysis explains how the comparison should be interpreted.
Does a higher score prove that an improvement worked?
No. A score change is evidence to investigate. Check the underlying issue, implementation, audience and other influences before making a causal claim.
Choose your next step
King Center
Related practice: reviewing participant feedback for program decisions. This story is not an industry NPS benchmark.
Read the customer story →