An NPS detractor is any customer who scores your business between zero and six on the likelihood-to-recommend question. That definition is simple and it is also where most NPS programs stop thinking, which is the problem. Seven of the eleven points on the scale fall inside the detractor band, so the label covers everyone from a customer who is furious about a botched delivery to a customer who is broadly content but never gives anyone a nine.
Those two people need completely different handling, and one of them may not need handling at all. Teams that respond to every detractor with the same apologetic follow-up spend their recovery capacity evenly across a group that is not evenly worth recovering, then conclude that detractor outreach does not work.
This guide covers what a detractor actually is, how the group differs from passives and promoters, how to identify detractors in your own data, the distinct types you will find once you look, how to respond to each, and how to move the ones who are reachable into the promoter band.
What is a Detractor in NPS?
A detractor is a survey respondent who answers the standard Net Promoter Score question with a score of six or lower on the zero-to-ten scale. In the calculation, detractors are the group subtracted from promoters, which means every detractor costs you twice: once in their own dissatisfaction and once in the arithmetic.
The reasoning behind the cutoff is behavioral rather than statistical. The band was set on the observation that customers who rate below seven do not reliably repurchase and are more likely to discourage others than to recommend, which makes them a churn and reputation signal rather than a satisfaction signal. That distinction matters because a detractor is not simply an unhappy customer. Someone can be satisfied with your product and still refuse to recommend it, usually because of price, category skepticism, or an unwillingness to attach their own reputation to a vendor.
The other thing the definition hides is intensity. A zero and a six sit in the same bucket and behave nothing alike. A six is a disappointed customer who is still engaged and still answering your surveys. A zero is often a customer already halfway out the door, and sometimes one who has already told others. Any program that reports a single detractor count without the internal distribution is discarding the information that decides who to contact first. The full definitional treatment of NPS covers the scale design in more depth.
Promoters vs. Passives vs. Detractors: What’s the Difference?
| Dimension | Detractors (0–6) | Passives (7–8) | Promoters (9–10) |
|---|---|---|---|
| Disposition | Dissatisfied or unwilling to endorse | Satisfied but unenthusiastic | Loyal and willing to advocate |
| Effect on the score | Subtracted from the total | Counted in the base, excluded from the score | Added to the total |
| Churn risk | Highest, and often nearest term | Moderate and quiet, since they leave without complaint | Lowest |
| Reputation effect | Can generate negative reviews and word of mouth | Neutral, contributes nothing either way | Generates referrals, reviews, and testimonials |
| Feedback volume | Highest, and the most specific | Lowest, frequently no comment at all | High, but usually general praise |
| What they tell you | Where the experience is failing and why | Almost nothing, which is their main risk | What is working and worth protecting |
| Right response | Service recovery, routed within 48 hours | Targeted improvement to the one blocking issue | Activation toward reviews and referrals |
| Owner | Customer success or service, as a work queue | Product or account management, as a campaign | Marketing, as an advocacy program |
| Score sensitivity | One shift up moves your NPS immediately | Movement in either direction changes the score | Already at the ceiling |
The strategic read is that detractors are the most expensive group and the most informative one, while passives are the most dangerous because they are invisible in both the score and the comment field. A base drifting into passives can leave your NPS flat while your risk profile deteriorates. Our breakdown of promoters, passives, and detractors goes further into how each group behaves over time.
How is NPS Calculated?
- Step 1: Ask the standard zero-to-ten recommendation question. Keep the wording and scale fixed so results stay comparable across waves.
- Step 2: Sort responses into the three bands. Zero through six are detractors, seven and eight are passives, nine and ten are promoters.
- Step 3: Convert detractors and promoters to percentages of total responses. Passives stay in the denominator. Skipped responses are excluded from the base entirely.
- Step 4: Subtract the detractor percentage from the promoter percentage. With 200 responses split into 100 promoters, 60 passives, and 40 detractors, that is 50 percent minus 20 percent, for an NPS of 30.
- Step 5: Report the result as a whole number between -100 and +100. An NPS of 30 is not 30 percent.
- Step 6: Report the confidence interval alongside it. NPS is a difference between two proportions, which makes it more volatile at low response counts than teams expect, and the margin of error calculator will size the interval for your response count.
The detractor percentage is worth tracking as its own line rather than only as an input to the score. A rising promoter count can mask a rising detractor count in the same wave, and the composite will move very little while both tails grow. A prebuilt NPS survey template reports the full distribution alongside the score by default.
How to Identify NPS Detractors?
- Step 1: Run the survey identified rather than anonymous where you can. Anonymous NPS produces a number and no work queue. If you need anonymity for candor, accept that recovery is off the table for that program and run a separate identified track for accounts.
- Step 2: Flag every response of six or below at submission. Not at the end of the wave. Detection needs to be automatic, which is what alerts and action plans are for.
- Step 3: Read the delta, not just the absolute score. A customer who dropped from nine to six is a different and more urgent signal than a customer who has scored six every quarter for two years. The first is a new problem, the second is a scale habit or a structural mismatch.
- Step 4: Segment detractors before contacting anyone. By account value, tenure, product line, region, and support history. Detractor lists sorted by account value get worked. Alphabetical lists do not.
- Step 5: Read the open text for the reason, then code it. The score tells you who, the comment tells you why, and theme frequency across comments tells you what to fix at the process level. Design guidance sits in our primer on open-ended questions.
- Step 6: Join the detractor list to your operational data. Support ticket counts, recent incidents, usage decline, and billing events usually explain the score before the customer does.
- Step 7: Look for detractor clusters, not just detractor individuals. Five detractors in one region or one product line in the same wave is a process failure. Five scattered across the base is normal variance.
→ Want detractors flagged and routed the moment they respond? Explore Sogolytics NPS software
Types of NPS Detractors
- Single-incident detractors. One bad experience against an otherwise good history, usually with a previously higher score. The most recoverable type and the best return on outreach.
- Chronic-friction detractors. Repeatedly low scores citing the same theme, such as slow support or a clumsy workflow. Outreach helps the individual; only a process fix helps the pattern.
- Expectation-mismatch detractors. Scores that fall after onboarding, when the product turned out to be something other than what was sold. The fix sits upstream in sales and onboarding, not in service.
- Price-value detractors. Content with the product, unwilling to endorse it at the price. Reframing value can work. Discounting usually converts them into a passive rather than a promoter.
- Feature-gap detractors. Blocked by something specific that does not exist yet. These are the most useful detractors you have, because they arrive with a prioritized roadmap request attached.
- Wrong-fit detractors. Customers who were never right for the product. Recovery effort here is close to unrecoverable, and a graceful exit costs less than a retention campaign.
- Proxy detractors. The respondent is not the person experiencing the product, common in B2B where a procurement contact scores on behalf of end users. Verify who answered before acting.
- Silent detractors. A low score and no comment. Common, and the reason a second follow-up channel matters.
- Score-artifact detractors. Respondents who never use the top of any scale, a pattern that varies considerably by region and culture. Their six may be someone else’s nine, and reading it as dissatisfaction generates outreach that confuses the customer.
The two types worth calling out are wrong-fit and score-artifact, because both look identical to a genuine detractor in a dashboard and neither responds to recovery. Filtering them out is what makes the remaining list workable.
Why Do Detractors Matter?
- They are your nearest-term churn risk. A detractor score is often the earliest named warning you get on an account, arriving before any usage or billing signal.
- They are the most specific source of diagnostic feedback. Dissatisfied customers explain themselves in detail. Promoters mostly say “great service.”
- They carry disproportionate reputation weight. Negative experiences travel further than positive ones, and a detractor who leaves a review does so where prospects are actively looking.
- They move the score twice. Every detractor converted to a promoter both removes a subtraction and adds an addition, which is why detractor work moves NPS faster than promoter work.
- They reveal process failures rather than individual ones. Clustered detractors point at a workflow, a region, or a handoff that no individual apology will fix.
- They predict revenue at risk in a way finance recognizes. Detractor count weighted by account value is a number a CFO will read.
- They test whether your program is real. A detractor who reports a problem and hears nothing back learns that the survey was decorative, and stops answering.
Working detractors as a routed queue rather than a report is what produces measurable movement. All In Credit Union structured member feedback around the moments that shape loyalty and pushed responses into service recovery in real time, lifting its Net Promoter Score by 20 points and reaching a 55 percent close rate on loan leads. The survey was the easy part. The routing was the achievement.
→ See the recovery workflow behind a 20-point lift. Read the All In Credit Union case study
How to Respond to NPS Detractors?
- Step 1: Alert the owner within the hour. Automatically, to a named person with an escalation path defined before launch. A daily digest is already too slow for a score of two.
- Step 2: Read the comment and the account history before making contact. Arriving without context and asking the customer to re-explain the problem makes the second impression worse than the first.
- Step 3: Reach out within 48 hours, by a human, on a channel they use. Recovery likelihood falls sharply after the first two days. A phone call for high-value accounts, email for the rest, and never a templated apology that reads as automated.
- Step 4: Acknowledge the specific issue rather than apologizing generally. Name what went wrong. Generic regret signals that nobody read the response.
- Step 5: Say what you will do and by when, or say plainly that you cannot fix it. Both are acceptable. A vague commitment is not, and an over-promise creates a second detractor event.
- Step 6: Route the underlying issue to the process owner separately. The customer conversation and the process fix are two different pieces of work, and doing only the first guarantees the same detractor next quarter. This is the difference closed-loop feedback is built around.
- Step 7: Log the outcome against the response. Resolved, unresolved, or declined contact. Without this you cannot report recovery rates, and recovery rate is the metric that justifies the program.
- Step 8: Re-survey in the next wave. Confirming the score moved is what turns an anecdote into evidence.
Two situations call for restraint. If the survey was anonymous, do not attempt identification, and instead publish what changed as a result of the wave so respondents see the loop close at a program level. And if a detractor asks not to be contacted, honor it immediately, since a second unwanted outreach converts a recoverable customer into a public one.
How to Turn Detractors Into Promoters?
- Step 1: Filter to the reachable ones first. Single-incident, chronic-friction, and feature-gap detractors are the working list. Wrong-fit and score-artifact detractors are not, and including them depresses your measured recovery rate.
- Step 2: Resolve the specific complaint completely before asking for anything. No survey, no upsell, no review request until the original issue is closed. Recovery has to land before advocacy is plausible.
- Step 3: Exceed the resolution slightly. Handling a problem better than the customer expected is one of the few reliable routes from detractor to promoter, and it works because it replaces the memory of the failure with the memory of the fix.
- Step 4: Fix the process behind the complaint and tell them you did. Naming the change the customer’s feedback produced is the single most powerful step in this sequence and the one teams skip most often.
- Step 5: Give the relationship time before re-measuring. A promoter score days after a resolution is a courtesy score. Re-survey on your normal cadence, or at least a month out.
- Step 6: Track recovery as a rate, by detractor type. Single-incident detractors will recover at multiples of the rate you see for price-value ones, and reporting a blended figure hides which recovery work is worth funding.
- Step 7: Escalate the pattern, not just the case. When the same theme recovers repeatedly through manual effort, the manual effort is subsidizing a broken process. Route the theme into the improvement backlog with the frequency data attached, which is what customer analytics reporting is for.
A realistic expectation helps here. Moving a detractor to a passive is a normal and valuable outcome, since it removes a subtraction from your score and a churn risk from your base. Full conversion to nine or ten is less common and should not be the benchmark you judge the program against. Teams starting from a lighter setup can run this workflow with a prebuilt instrument from the survey template library, and the metric mix of NPS, CSAT, and CES determines how early you catch the friction that creates detractors in the first place.
Conclusion
An NPS detractor is a score of six or below, but the useful work starts after that definition: separating the recoverable types from the ones that are not, routing them fast enough to matter, and fixing the process behind the pattern rather than the case in front of you.
Sogolytics supports that loop end to end, from distribution and voice of the customer collection through sentiment analysis on open text, real-time alerting, and recovery tracking inside the customer experience platform.
→ See what detractor recovery looks like with routing built in. Browse customer stories
FAQs About NPS Detractors
What score makes someone an NPS detractor?
Any score from zero through six on the zero-to-ten recommendation scale. That covers seven of the eleven available points, which is why the group is so internally varied. Scores of seven and eight are passives, and nine and ten are promoters, so a detractor is not simply someone who scored below average but someone whose score falls in the band associated with non-repurchase and negative word of mouth.
How quickly should you respond to a detractor?
Alert the owner within the hour and make human contact inside 48 hours. Recovery likelihood declines sharply after the first two days, since the customer has by then either resolved the issue themselves, told someone about it, or started evaluating alternatives. For high-value accounts and for scores of three or below, same-day contact is the right target.
What if a detractor leaves no feedback?
Silent detractors are common and still worth contacting, since the absence of a comment is not the absence of a reason. Check operational data first, including recent support tickets, incidents, usage decline, and billing changes, which frequently explains the score without the customer having to. Then reach out with a specific question about what you found rather than a general request for feedback, because an open-ended ask has already failed once.
How does closed-loop feedback help with detractors?
Closed-loop feedback turns a detractor score into an assigned task with a deadline instead of a line in a report. It covers the alert, the owner, the outreach, the logged outcome, and the process fix routed separately from the customer conversation. Programs without it show no measurable retention effect from NPS, which is the strongest available argument that the loop, not the score, is what produces the return.
Can you save every detractor?
No, and building the program on that assumption wastes recovery capacity. Wrong-fit customers who were never suited to the product will not be retained by better service, and score-artifact respondents who never use the top of a scale were not dissatisfied to begin with. Filtering both out concentrates effort on single-incident, chronic-friction, and feature-gap detractors, where recovery rates are high enough to justify the cost.
Can every NPS detractor become a promoter?
Full conversion to a nine or ten is realistic only for a subset, typically single-incident detractors whose problem was resolved better than they expected. For most of the reachable list, the achievable outcome is a move into the passive band, which still removes a subtraction from your score and a churn risk from your base. Judging the program on promoter conversion alone understates its actual return, so track movement out of the detractor band as the primary measure.


