What the Net Promoter Formula Actually Measures
Net Promoter Score compresses survey data into one number: the percentage of promoters minus the percentage of detractors. Respondents who answer the recommendation question 9 or 10 are promoters, 7 or 8 are passives, and 0 through 6 are detractors. The result sits on a scale from −100 to +100, where zero means promoter and detractor shares cancel exactly. Ten promoters, ten passives, and ten detractors is the canonical flat result — 33.3% on each side and a score of 0.
The subtraction is what separates this metric from an average rating. A company where 45% of customers are promoters and 40% are detractors looks acceptable on a mean score near 7 out of 10, yet the NPS of +5 exposes the problem. The metric treats a vocal unhappy customer as an active drag on growth, not merely a low data point pulling down an average.
Percentages, not raw counts, drive the math, so survey size does not distort comparisons between waves or segments. The default numbers above show the arithmetic: 120 promoters and 20 detractors across 200 responses give 60% versus 10%, a score of +50. Raw counts from the same survey would mean nothing without the denominator behind them.
The 0-10 Question and Where the Lines Fall
The survey asks one question: 'How likely is it that you would recommend this company to a friend or colleague?' on a 0-10 scale. The boundaries are fixed by convention — 9-10 promoter, 7-8 passive, 0-6 detractor — and they are not symmetric. Someone answering 6 is grouped with someone answering 0, because the research behind the metric found 7s behave closer to satisfied-but-silent customers than to angry ones.
The asymmetry surprises people, but it reflects observed behavior: detractors at 0-6 actively discourage others, passives at 7-8 are content but switchable, and only 9-10 respondents reliably refer. Changing the cutoffs — counting 8 as a promoter, say — breaks comparability with every published benchmark and with your own history. Keep the lines where the convention puts them.
Pair the scale question with one open text field: 'What is the most important reason for your score?' The number segments the base; the verbatim explains it. Most survey tools export both, and the counts you paste into this calculator come straight from the cross-tab of that scale question.
Reading Your Score Against the Bands
The common bands: below 0 is concerning, 0-30 needs improvement, 30-50 is good, 50-70 is excellent, and 70+ is world class. The default scenario lands at +50, the edge of excellent. These bands are heuristics rather than standards — no governing body defines them — but they give leadership a shared vocabulary for where the number sits.
Movement matters more than the absolute level. A team climbing from +12 to +31 over three quarters has usually fixed something real, even if both numbers sit inside average territory. Sharp drops of ten points or more deserve investigation before the quarter-over-quarter reporting reaches stakeholders.
Track the score next to your retention numbers — the customer retention rate calculator makes the pairing easy — because willingness to recommend usually decays before a subscription actually lapses. A slipping NPS paired with still-flat retention is an early-warning combination worth watching closely.
Passives and the Denominator Effect
Passives appear in the denominator but on neither side of the subtraction, so a fat middle quietly drags the score toward zero. A 1,000-response survey with 200 promoters, 700 passives, and 100 detractors scores +10, even though promoters outnumber detractors two to one. The same promoter and detractor counts with zero passives would score +33.3 instead.
This is a feature, not a bug. Passives are satisfied but unenthusiastic — they bought from you and might buy again, but they also switch for a marginally better price. A swelling passive share tells you the experience is acceptable without being remarkable, which is exactly the signal the subtraction is built to surface.
Teams often try to convert passives into promoters by over-delivering on the one dimension their verbatims mention most. That is the correct play: detractor recovery stops the bleeding, but passive conversion is usually where the cheapest score gains hide, because passives already like you enough to stay.
Response Counts, Swings, and Statistical Noise
Every response carries weight equal to 100 divided by the sample size. At 200 responses, each person moves the score 0.5 points, and one respondent switching between promoter and detractor swings it a full point. At 50 responses, a single person is worth 2 points — enough to turn an ordinary wave into a fake record.
Statistical noise compounds the swing. A 200-response survey carries a worst-case 95% margin of roughly ±14 points on the score itself, so treat single-digit gaps between waves or competitors as ties. Below about 100 responses, the number is directionally useful but hard to quote credibly.
Small samples also produce absurd results — nine promoters and one passive in a ten-person pilot scores +90. Resist publishing that. Wait for volume to build, and when sample sizes differ between segments you are comparing, weight or resample them to comparable counts first.
Industry Benchmarks in Context
The benchmark menu uses approximate midpoints of ranges published in industry studies: SaaS and software typically lands in the +30s, retail and e-commerce in the +40s, financial services in the high +30s, healthcare in the mid +40s, and telecom in single digits. Treat every published figure as directional — methodologies, sampling, and wording vary between studies and shift a few points year to year.
The gap to the benchmark, not the raw score, is the takeaway. The default +50 sits +18 above a cross-industry midpoint of roughly +32, +14 above the SaaS figure, and only +5 above healthcare. The same score can read as leading, average, or trailing purely because of the industry you picked.
Two practical rules keep benchmarking honest: compare like survey types (relationship surveys with relationship surveys), and re-verify the benchmark against a current source before quoting it in board material. A telecom operator celebrating +15 against a +44 retail figure is celebrating the wrong thing.
Turning Detractors Around
Detractor recovery is operational, not analytical: contact every 0-6 respondent within 48 hours, ask what went wrong, and fix what you can. Teams that close the loop consistently convert a slice of detractors into promoters over time. In the default 200-response pool, converting ten detractors into promoters moves the score from +50 to +60 — a ten-point gain from 5% of the base.
Track the promoter-to-detractor ratio alongside the score; the default split is 6 to 1. Ratios below 2 to 1 usually mark real churn risk even when the headline number looks passable, because a large detractor block actively spreads negative word of mouth between your customers and their peers.
Score improvements mean little if customers keep leaving, so read NPS next to your churn numbers. The churn rate calculator turns subscription cancels into the monthly and annualized rates that tell you whether the recovery work is actually sticking.
NPS and Unit Economics
Promoters are worth more because they stay longer, buy more, and refer — three behaviors that compound. Fred Reichheld, the metric's creator, reported that long-term growth leaders typically outgrow competitors by roughly a factor of two, with high promoter share cited as a leading driver. Verify current studies before quoting that figure, but the direction is well documented.
Retention gains flow straight into lifetime value, so model them explicitly: the CLTV calculator shows how a few extra months of average tenure change lifetime value, and the CAC calculator prices what it costs to replace a defected customer with a new one. When retention rises, the ratio between those two numbers moves in your favor without touching ad spend.
That is also how to frame the score for finance: as a leading indicator feeding the ROI calculator inputs rather than a vanity metric. A retention program that lifts NPS by ten points and cuts churn by a point or two pays for itself quickly in most subscription businesses, and the calculator makes that case with your own numbers.
Survey Design Mistakes That Skew the Score
Wording decides the score. Leading phrasing such as 'How much do you love our product,' branded sender names, and questions placed right after a discount all inflate promoter counts. Keep the question verbatim, send from a neutral address, and avoid running the survey the same week as a price increase or a public outage.
Sampling bias is deadlier than bad wording. Surveying only customers who just contacted support, only app users, or only recent signups produces a score that describes a slice, not the base. For email-delivered surveys, the invitation click-through shapes who answers at all — the CTR calculator helps you check whether the responding sample represents everyone you invited.
Incentives are the quiet score-killer. A prize draw converts annoyed customers into polite 8s and motivates freebie-hunters to click 9, which raises the number while destroying its meaning. Run clean surveys, accept slightly lower response rates, and keep waves comparable to each other.
Employee Net Promoter Score (eNPS)
The same machinery runs on employees: 'How likely are you to recommend this company as a place to work?' on the same 0-10 scale, with identical 9-10 / 7-8 / 0-6 boundaries. A workforce of 45% promoters, 35% passives, and 20% detractors scores +25. Anonymous collection is non-negotiable — attributed eNPS surveys reliably measure fear rather than loyalty.
eNPS trades the customer-churn link for an attrition one. Low scores tend to run ahead of regretted departures, and teams that interview leavers often find they answered as passive or detractor on the last internal survey. The attrition rate calculator turns departures into the rate that should move opposite to your eNPS trend.
Keep employee and customer programs separate in reporting. The numbers look identical on a chart but answer different questions, and blending them invites bogus comparisons. Run eNPS twice a year at most — more frequent pulses fit better on a five-point engagement scale than on the 0-10 recommend question.