Create NPS survey — measure Net Promoter Score

The Net Promoter Score (NPS) is based on a single question: "How likely are you to recommend us?" — on a scale of 0 to 10. Answers are segmented into three groups: 0–6 are detractors (dissatisfied), 7–8 passives (neutral), 9–10 promoters (enthusiastic).

The calculation itself is simple: percent promoters minus percent detractors. Passives are deliberately ignored as they make neither positive nor negative recommendations. Example: 60 percent promoters, 20 percent passives, 20 percent detractors yields an NPS of +40. The score ranges between -100 and +100, where anything above 0 is considered good, above 30 very good, above 50 excellent.

Important to understand: NPS is a relative metric, not an absolute scale. Industries differ widely — SaaS providers often reach +30 to +50, telecom providers more like -10 to +10. Compare yourself to your industry, not to Apple. And track the trend over time — a single measurement is gut feeling, a quarterly progression is information.

An NPS survey without follow-up action is wasted data collection. Whoever gives 3 points has a concrete problem — and right now, not in three months at the quarterly report. Conditional logic enables a differentiated reaction directly in the form.

Proven pattern: at score 0–6 a follow-up question appears "what went wrong?" as an optional text field. At 7–8 you ask "what would move you to a 9 or 10?". At 9–10 ask for a referral or testimonial. Three different paths, one form — the respondent does not notice the routing.

For quick reaction, an automatic webhook on detractors pays off additionally: Slack notification to the customer success team, automatic ticket creation in the helpdesk, or an email to the responsible account manager. Anyone contacting a detractor personally within 24 hours can save the relationship in many cases — and gains valuable insights why things went wrong on top.

Anyone surveying customers too often gets no answers in the end. Survey fatigue is real — studies show response rates drop noticeably after three to four surveys per year, often below 5 percent. A sensible cadence is therefore not a question of "more is better" but of discipline.

For transactional NPS (after a concrete event: support ticket, purchase, onboarding completion) one survey per event is okay — the context is fresh. For relationship NPS (general satisfaction) once per quarter suffices, in B2B often just twice a year. Anyone asking monthly gets ignored.

The time of day also matters. B2B surveys work best Tuesday to Thursday mornings — Monday gets ignored, Friday is mentally already weekend. B2C surveys run better evenings and weekends. A short subject line without marketing language raises open rates: "One question about your experience" beats "We would love your feedback!".

Detractor answers need attention — and not just at the next reporting. An automatic alert in Slack or Microsoft Teams ensures the right person immediately knows what happened and can react.

The mechanic: a webhook forwards every answer with score ≤ 6 to a dedicated channel — including customer name, score and free-text reason. The customer success team sees the entry the moment the answer arrives and can contact the customer the same day. That is the difference between "complaint heard" and "customer lost".

For aggregate evaluations, a connection to a BI tool like Metabase, Looker Studio or a simple Google Sheets pipeline pays off. You see trends per segment (customer cohort, product version, region) and can adjust specifically. Important: alert only on real issues, otherwise alert fatigue arises. Three detractors per day are worth an alarm — three per week are more statistics.

A single NPS value is nearly useless. Only the comparison over time shows whether you are getting better or worse. Whoever measures NPS +35 in Q1 and +42 in Q2 did something right — and should find out what. The absolute number matters less than the direction.

For a clean trend analysis you need three things: consistent methodology (always the same scale, the same question, the same target group), sufficient sample size (at least 100 answers per period for statistical significance) and segmentation (B2B vs B2C, new vs existing customers, product versions).

Beware of survivor bias: anyone asking only satisfied customers — because dissatisfied ones already churned — gets a beautified score. A healthy NPS program also measures cohorts shortly before contract end or after support escalation, not only the loyal promoters. Connection to the CRM with hidden fields (contract duration, plan, onboarding date) enables later evaluation by segment — without this metadata the NPS is just a one-dimensional number.