What is an NPS detractor?
An NPS detractor is a respondent who gives a recommendation rating from 0 to 6 on the 0–10 Net Promoter Score question. Ratings of 7–8 are passives and 9–10 are promoters. These categories describe the response; a low rating alone does not prove that a customer will leave or discourage others.
The useful next step is to understand what the respondent is telling you. A comment may identify a service problem, an unmet expectation or a reason the recommendation question did not fit their situation. Some respondents give no explanation. Do not invent one from the score.
This guide shows how detractors affect the calculation, how to review their feedback, how to choose a response and how to check what happened afterward. The goal is a better customer experience, rather than persuading someone to give a higher number.
How do detractors affect NPS?
NPS equals the percentage of promoters minus the percentage of detractors, using all valid ratings as the denominator. Passives are included in that denominator. See Bain’s scoring guidance.
Suppose a fictional survey has 100 valid responses: 50 promoters, 30 passives and 20 detractors. NPS is 50 − 20 = 30. The detractor rate is 20/100, or 20%. It is related to NPS but is not the same measure.
With that denominator held fixed, changing one rating from passive to detractor lowers NPS by one point. Changing one rating from promoter to detractor lowers it by two points because one promoter is lost and one detractor is added. It is therefore inaccurate to say that every detractor always has “twice the effect of a passive.” State what changed and keep the denominator visible.
If the next wave has a different number or mix of respondents, recalculate from that wave’s counts. Do not interpret a change in the aggregate score as if the same people necessarily changed their views.
What a low recommendation rating can and cannot tell you
A low rating is a useful reason to look more closely. It is not a diagnosis. Two people who both choose 4 may be describing very different experiences: one could have an unresolved account issue, while another likes the service but rarely recommends products to others.
Read the explanation alongside the rating and relevant context. Where appropriate, that may include the product, touchpoint, account role and date. Use only context the team is authorized to access. Do not assume a single comment captures the customer’s entire relationship.
Likewise, a promoter can mention a serious problem and a passive can offer a practical improvement. Review the whole feedback set while keeping the detractor view available. Restricting attention to low scores can hide issues elsewhere.
Review detractor comments in five steps
- Confirm the dataset. Check the rating range, collection period, duplicates and exclusions.
- Keep missing explanations visible. Count detractors with and without usable comments separately.
- Code the issues. Use clear topic definitions and allow more than one topic where needed.
- Check source passages. Review the evidence behind each theme, including exceptions and contradictions.
- Separate findings from actions. Describe what was reported before deciding what the organization should do.
For example, “billing” may be too broad to assign an action. Distinguish an unclear invoice, an unexpected charge and a failed payment process when the comments support those differences. Do not create false precision when the response only says “billing is bad.”
Automated text analysis can help organize a large set of comments. A reviewer still needs to check whether the labels fit and whether a summary overlooks important differences. Keep the coding definitions and corrections so the next reporting period can be compared sensibly.
Worked example: turn feedback into a review list
Imagine 20 detractors in the 100-response survey. Fifteen provide usable comments. Eight mention onboarding delays, four mention billing confusion and two mention device reliability. Some comments have multiple topics, and some describe other issues. These are fictional figures used to illustrate the method.
Scroll horizontally to see all columns →
| Reported issue | Evidence | Question for the owner |
|---|---|---|
| Onboarding delays | Eight of 15 detractor comments mention the issue. | Which step and period do these comments describe? |
| Billing confusion | Four comments describe difficulty understanding charges. | Is the charge wrong, the explanation unclear, or both? |
| Device reliability | Two comments report failures. | Is there a time-sensitive service or safety issue? |
Eight of 15 is about 53.3% of detractor comments, not 53.3% of all customers. Eight of 20 is 40% of detractor respondents. State the chosen denominator. The five detractors without usable comments remain an information gap.
The smaller reliability theme may need attention before the larger onboarding theme if the consequences are more serious. Frequency and average score help describe the data; they are not a complete prioritization rule.
How should you prioritize detractor issues?
Consider the nature and severity of the issue, how many people appear affected, whether the problem is continuing and what evidence needs verification. Then consider the owner’s ability to act and the likely benefit of the proposed change.
Do not automatically rank a frequent inconvenience above a rare but serious problem. Equally, one strongly worded comment should not displace a well-supported recurring issue solely because it is vivid. Review the source and operational context before deciding.
A practical action register contains the issue, supporting evidence, relevant period, affected group, owner, next step, due date and review status. Record uncertainty where the team still needs to investigate. Calling a theme a “driver” should not imply that a causal analysis has established it.
Separate the response to one person from the process improvement. Correcting an individual invoice may help that customer immediately. Changing an unclear billing explanation may help others too. These actions can be related without being interchangeable.
Respond to the issue, not the score
Where named contact is appropriate, acknowledge the concern and explain what the team can do next. Refer to the actual issue rather than beginning with a demand to justify the rating. Do not pressure the respondent to change a score or make support conditional on doing so.
A simple response might read: “Thank you for explaining the difficulty with setup. I’d like to check which step failed and help resolve it. Is this a suitable way to contact you?” Adapt the wording to what the customer has shared and the contact arrangements they agreed to.
If the survey is anonymous, do not attempt to identify the person from clues in their comment. Use an appropriate public or group update about the pattern, or offer a voluntary support route. Anonymous feedback can lead to action even without a named personal record.
Not every low score requires the same response. Follow the applicable support process for an urgent issue, investigate unclear reports and acknowledge themes through the right channel. Avoid promising a fixed resolution date before the responsible team can commit to it.
How do you know whether the response helped?
Record the action taken and the evidence of resolution. A sent email is evidence of contact, not proof that the problem was solved. Useful checks may include the customer’s confirmation, a corrected transaction, completion of the blocked task or an operational measure that reflects the issue.
A later recommendation rating can add context, but it is not the only outcome. The customer may have a resolved issue and still be unlikely to recommend the service. They may also not respond to the next survey. Keep those possibilities separate from failure or success.
Where repeated named responses are collected appropriately, show the matched group separately from the full wave. A rise in one person’s rating describes a change; it does not prove your action caused it. For a broader process improvement, examine comparable periods, collection conditions and other changes that could affect the result.
Keep the feedback process consistent across teams
Use a small common set of definitions for score groups, dates, issue categories and action status. Different products or locations can retain local topics. Document how local categories map to common themes and avoid combining issues that merely share a similar label.
Preserve the original comment and the reviewed interpretation. Record a definition change so a rise in theme frequency can be examined against the change in coding. Limit access to personal feedback and account notes according to the work people need to perform.
Sopact’s relevant role is connecting collection, analysis and governance so an operational team can maintain this evidence and its review. Test your proposed setup with ambiguous comments, missing follow-up and a changed definition. A connected record makes the history easier to inspect; it does not make every interpretation correct.
Build the wider customer feedback workflow
Use the NPS software buying checklist to evaluate tools against a complete collection and follow-up cycle. Use NPS survey questions to improve how you ask for the initial explanation.
For a program that combines several sources and teams, continue to customer feedback management. Keep the practical course linked below as the path for turning these ideas into a repeatable workflow.
Watch: reviewing incoming feedback
This video discusses reading feedback as it arrives and retaining the context needed for review. Apply the approach with the identity and privacy arrangement appropriate to your survey.
Frequently asked questions
Is a rating of 6 an NPS detractor?
Yes. Detractors give ratings from 0 through 6. Passives give 7 or 8, and promoters give 9 or 10.
Are passives ignored in NPS?
No. They remain in the valid-response denominator. The formula subtracts the detractor percentage from the promoter percentage without removing passive responses first.
Will every detractor leave?
No. The score does not establish an individual customer’s future behavior. Review the explanation and relevant context instead of treating the category as a churn prediction.
Should the most frequent issue always be fixed first?
No. Frequency is one consideration. Severity, urgency, the evidence and the consequences for affected people also matter. A rare serious issue may warrant immediate attention.
Does a higher follow-up score prove the fix worked?
No. It is one piece of evidence. Check the actual issue, action and resolution, and consider other influences on the later rating.
Can anonymous detractor feedback lead to action?
Yes. Teams can investigate patterns and improve processes without identifying individuals. Preserve the promised anonymity and offer voluntary support options where appropriate.
Choose your next step
King Center
Related practice: using participant feedback to inform program review. This story is not an NPS outcome claim.
Read the customer story →