How to calculate your NPS in 5 steps — and what the score really reveals

NPS: what that single question actually measures

Behind its apparent simplicity — one question, a score between −100 and +100 — the Net Promoter Score is one of the richest indicators available to an SMB. Developed by Fred Reichheld and Bain & Company, it rests on a premise validated by decades of research: the propensity to recommend is the best predictor of organic growth.

But calculating an NPS and drawing actionable decisions from it are two very different exercises. Here are the five steps to do both correctly.

Step 1 — Wording the question correctly

The standard NPS question is non-negotiable in its wording. Any modification — even a minor one — makes your score incomparable to industry benchmarks and your own historical data.

Relational vs transactional context

There are two NPS delivery modes, each with distinct uses:

• Relational NPS: sent on a fixed cadence (quarterly or semi-annually) to your entire customer base — measures the overall health of the relationship.
• Transactional NPS: triggered after a specific event (renewal, delivery, onboarding) — measures the impact of a particular moment on loyalty.

For an SMB launching its first programme, relational NPS is the priority. Transactional NPS comes as a complement once the first is well established.

Step 2 — Collecting responses rigorously

The quality of your NPS score depends directly on the quality of your data collection. Three parameters determine the reliability of your data.

Response rate

An NPS calculated from 12 responses has no statistical value. To be reliable, aim for a minimum of 30 to 50 responses per analysed segment. In practice, for SMBs:

• Average email response rate: 15% to 30% depending on the depth of the relationship
• In-app response rate: 20% to 40% when the trigger is contextually relevant
• A single follow-up after 5 days: increases the rate by an additional 20% to 35%

Send timing

Avoid sending a relational NPS within 30 days of a known bad experience (unresolved ticket, delivery delay). The score will be biased downward and will not reflect actual loyalty. Equally, do not send it right after an exceptional win — you will get a temporarily inflated score.

Representativeness

Make sure your sample covers all your segments: tenure (new vs long-standing clients), account size, industry. An overall NPS of 42 that hides an NPS of 12 among your recent SMB clients is a critical alert that only segmentation reveals.

Step 3 — Calculating the score

The calculation is straightforward. What matters is fully understanding the three segments.

Concrete example

Step 4 — Interpreting the score in context

Is an NPS of 28 good or bad? The answer depends entirely on context. Here are the three dimensions of interpretation that actually matter.

The trend over time

This is the most important dimension. An NPS of 28 progressing to 38 over four consecutive quarters is a strong traction signal. An NPS of 50 declining to 35 is a serious alert, even if the absolute score remains high. Always analyse the curve, never the isolated data point.

The industry benchmark

Scores vary enormously across industries — an NPS of 28 is above average for B2B SaaS services, but would be considered low in the insurance sector.

Internal segmentation

Always break down your overall NPS by meaningful segment: tenure (clients <6 months, 6–24 months, 24+ months), account size, product used, account manager. Gaps between segments pinpoint precisely where to focus improvement efforts.

Step 5 — Activate: turning the score into decisions

The fifth step is the one most SMBs skip — and that is precisely why their NPS stagnates. A score without an activation system is an academic exercise.

Detractor follow-up protocol (within 48 hours)

• Personalized acknowledgement within 24 business hours — not a generic automated email
• Understanding call from the account manager or a senior stakeholder
• Documented action plan shared with the client within 5 business days
• 30-day check-in to measure the evolution of perception

Promoter activation protocol

• Invitation to a structured referral or ambassador programme
• Request for a testimonial or case study (with explicit consent)
• Invitation to product betas or strategic advisory groups

Integration into leadership reviews

NPS must appear in every monthly or quarterly review with three elements: the current score, the trend over 4 periods, and the three most recurring verbatims (positive and negative). Without this integration, the score generates no organizational pressure to change.

The most common calculation and interpretation mistakes

1. Modifying the question. Adding "and to your family?", changing to a 5-point scale, rephrasing — any modification invalidates comparability with benchmarks and your own historical data.
2. Calculating from too small a sample. Below 30 respondents, the statistical margin of error exceeds 15 points. An NPS of 32 calculated from 18 responses is statistically non-significant.
3. Ignoring passives. Your 7s and 8s are promoters-in-waiting — or potential detractors. Analysing their verbatims often reveals exactly what is missing to cross the threshold to active recommendation.
4. Comparing across industries. An NPS of 40 in financial services is worth more than an NPS of 40 in retail — expectations and referral dynamics are fundamentally different.
5. Measuring without an open-ended question. The score tells you how many. Without the "why?" question, you do not know what to change. The open-ended question is mandatory.

To complete your measurement programme, discover how to choose between NPS, CSAT and CES for your context and when to combine all three metrics for a complete picture.