Definition (canonical)
NPS (Net Promoter Score) measures customer advocacy with one question — "how likely are you to recommend us, 0–10?" — and one formula: % promoters (9–10) minus % detractors (0–6). Scores range −100 to +100; passives (7–8) count in the denominator but neither add nor subtract.
Computing and reading it
Survey 100 customers: 45 promoters, 35 passives, 20 detractors → NPS = 45 − 20 = +25. Reading requires context on three axes: industry (telecoms live near zero; beloved consumer brands exceed +60 — cross-industry comparisons mislead); trend (your NPS versus last quarter beats any absolute); segment split (an average hiding +50 enterprise and −10 SMB is hiding your strategy question). The score's real predictive link runs through behavior: promoters churn less and refer more — which is why NPS correlates with retention when collected honestly.
Why NPS gets gamed — and how to keep it honest
The metric's fame made it a bonus target, and bonus targets corrupt collection: surveying only after happy interactions, staff begging for tens ("anything less than 9 counts against me"), excluding churned customers — each inflates the number while destroying its meaning. Honest protocol: fixed sampling independent of interaction outcome, third-party or automated delivery, churned customers included in relationship surveys, and the follow-up question ("why?") treated as the main product — the score is a headline; the verbatims are the content. Closing the loop with detractors within days converts measurement into retention work.
Where NPS fits in the metric stack
NPS is attitudinal — it asks about intent, not action. Pair it with behavior: churn and NRR tell you what customers DID; NPS explains the leading mood and flags risk before renewal dates. Two honest variants: relationship NPS (periodic, strategic) vs transactional NPS (post-interaction, operational) — mixing their numbers produces noise. And resist NPS-as-only-KPI: a company can buy +20 points with unsustainable generosity; the KPI stack needs a profitability counterweight.
FAQ
What is a "good" NPS? Above 0 = more advocates than critics; +30–50 strong in most B2B; +70 rare air. Your industry median and your trend are the real benchmarks.
Sample size needed? Enough for stable splits — as a rule of thumb 100+ responses per segment you intend to read; below that, watch verbatims, not the score.
NPS vs CSAT vs CES? CSAT rates satisfaction with a moment; CES rates effort; NPS approximates the relationship. Operational fixes → CSAT/CES; strategy → NPS.
Related: Retention · Churn · Customer Success · KPI Reviewed: 2026-07-15 · Sources: NPS canon