- What's the formula for NPS?
- NPS = (% promoters) − (% detractors). Promoters are respondents who scored 9 or 10. Detractors scored 0 through 6. Passives (7 or 8) are part of the total but contribute 0 to the calculation — they cancel out by design. The result lands somewhere between −100 (every respondent is a detractor) and +100 (every respondent is a promoter).
- Why do passives not count?
- Fred Reichheld, who introduced NPS at Bain & Company in 2003, framed it as a measure of growth fuel and growth drag. Promoters bring in new customers through word-of-mouth (fuel); detractors actively push them away (drag). Passives are happy enough to come back but won't recommend you and won't warn anyone off — they're neutral on growth, so they don't move the score. Counting them would dilute the signal.
- Is a 6 really a detractor? They didn't seem unhappy.
- Six is the most-litigated cutoff in NPS, and the answer is yes — but the framing matters. The original Bain research found that customers who scored 0–6 churned at much higher rates and recommended companies far less often than those who scored 7+. Calling them 'detractors' overstates how angry they are; what the number really means is 'won't help you grow.' Many of them are simply indifferent. Read the open-text comments to understand which ones are actually unhappy.
- What's a good NPS score?
- The widely-used Bain bands are: below 0 (needs work — more drag than fuel), 0 to 30 (good — most companies live here), 30 to 70 (great — strong word-of-mouth), and above 70 (world-class — Apple, Costco, Tesla territory). But industry context matters: an NPS of +30 is excellent for a bank and merely okay for a SaaS product. Use the benchmark table on this page as a starting point, not a verdict.
- How many responses do I need before the score is meaningful?
- There's no magic number, but the variance gets ugly below about 30 responses — a single detractor in a sample of 10 swings the score by 10 points, which is noise, not signal. For internal tracking, 100+ responses per period is a reasonable floor. For comparing segments (mobile vs desktop, free vs paid), you want at least 100 per segment. Below 30, treat the number as a directional read on a small sample, not a metric.
- Should I survey all customers or just recent ones?
- Two common methods. 'Relational NPS' surveys all customers on a fixed cadence (quarterly, biannually) and measures overall sentiment. 'Transactional NPS' surveys customers immediately after a specific interaction (a support ticket, an onboarding step, a checkout) and measures sentiment about that touchpoint. Most teams run both: relational for the headline trend, transactional for finding which moments break things. The calculator works the same for either — just pick the population you care about and survey them.
- What do I do once I have the score?
- The score is the headline; the work lives in the comments. Read every open-text response from your detractors — they'll usually tell you exactly what's wrong in plain language. Categorize the themes (pricing, support response time, missing feature, broken flow). Pick the one or two that come up most often, fix them, and re-survey in 60–90 days. NPS is most useful as a trend line, not as a single number — moving from +15 to +25 over two quarters is a real signal that the fixes worked.
- Does anything stored or sent anywhere when I use this calculator?
- No. The calculation runs entirely in your browser. The raw scores you paste, the counts you enter, the computed NPS — none of it leaves the page or gets logged. Close the tab and it's gone.
