How to Calculate NPS (and What to Do With the Score)

Here is how I calculate NPS and what I do with the score. Net Promoter Score (NPS) is the percentage of promoters (answers of 9 or 10) minus the percentage of detractors (answers of 0 to 6) on the question "How likely are you to recommend us to a friend or colleague?". NPS always lands between −100 and +100: 212 promoters and 80 detractors out of 400 answers give an NPS of 33.

If you only need the number, paste your three counts into the free NPS calculator. Below is how I calculate NPS by hand, how much I trust the result, and what I do once it comes in.

How do you calculate NPS?

NPS = (Promoters ÷ Total responses × 100) − (Detractors ÷ Total responses × 100)

Promoters        = answers of 9 or 10
Passives         = answers of 7 or 8 (counted in the total, not in the subtraction)
Detractors       = answers of 0 to 6
Total responses  = Promoters + Passives + Detractors

Two details catch people out. You subtract percentages of all responses, not raw counts. And the result is a plain number: an NPS of 33 is "33", not "33%".

What does an NPS calculation look like with real numbers?

When I ran support for hotel partners and travellers, we sent the NPS survey once a quarter. Here is one quarter, rounded:

Input	Count
Surveys sent	2,000
Responses	400
Promoters (9–10)	212
Passives (7–8)	108
Detractors (0–6)	80

Check the total first: 212 + 108 + 80 = 400. If your three groups do not add up to your responses, find the gap before you do anything else.

1. % promoters = 212 ÷ 400 × 100 = 53%
2. % detractors = 80 ÷ 400 × 100 = 20%
3. NPS = 53 − 20 = 33

Response rate = 400 ÷ 2,000 × 100 = 20%. You can check it with the survey response rate calculator.

Run your own counts through the NPS calculator and compare them with this example.

Why can't I just average the 0 to 10 scores?

Because an average hides the split that NPS is built to show. Imagine 400 customers who all answer 8. The average is 8.0, which sounds fine. The NPS is 0, because every answer is a passive. Now imagine half answer 10 and half answer 6. The average is still 8.0, but the NPS is also 0, and you have 200 unhappy customers to call. Same average, very different quarter.

How much can I trust my NPS?

Less than most dashboards suggest. NPS is the difference of two percentages, so it moves more than a single percentage does. With the numbers above (53% promoters, 20% detractors, 400 answers), the 95% margin of error is roughly ±8 points. With 100 answers in a segment it grows to about ±15 points.

So when a segment of 60 partners drops 10 points, I read the comments before I call a meeting. When the full sample of 400 drops 11 points, I treat it as real. The survey sample size calculator shows how many answers you need for the precision you want.

What is a good NPS score?

It depends on your industry and on who you survey, so I don't copy outside benchmarks. A score from a relationship survey sent to all partners is not comparable with one sent right after a solved ticket.

These are the working thresholds I used for a B2B travel support team:

• Below 0: more detractors than promoters. Act this week.
• 0 to 30: normal for us, with clear themes to fix.
• Above 30: a good quarter, as long as detractors stay under about 20%.

Your own trend, on the same question and the same audience, tells you more than any table.

What I've seen in practice

The score on its own does nothing. These are the steps I follow every quarter:

• Call every detractor who left a comment within 48 hours. Not an email. A call or a short video meeting.
• Split the score by segment. B2B partners and B2C travellers gave very different scores. One blended number hid both stories.
• Tag the comments. I group detractor comments into 5 to 8 themes (refunds, response time, booking errors) and count them.
• Fix one theme per quarter. One fix done well beats five half fixes.
• Ask promoters for something. A review, a case study, a referral. They already said yes in the survey.

The mistake I see most often is surveying only happy moments. If NPS goes out only after a solved ticket, the score looks better than reality, because the customers still waiting never get asked.

One quarter our NPS dropped 11 points. I expected a product issue. The comments said something else: travellers waited too long for refund updates. Nobody was angry about the refund itself, only about the silence. We added one proactive status email at day 3 and day 7. The next quarter the score came back, and refund tickets dropped too.

My NPS dashboard

I keep the same table every month. Because we survey quarterly, "this month" shows the latest wave and "last month" the wave before. These numbers are the recovery quarter after the refund fix.

Metric	Target	This month	Last month	Change
NPS	30+	33	22	+11
Promoters	50%+	53%	45%	+8 pts
Detractors	under 20%	20%	23%	−3 pts
Responses	400+	400	360	+40
Response rate	15%+	20%	18%	+2 pts

If you also run CSAT or CES, read how I choose between CSAT, NPS and CES, or see the full list of customer experience metrics. For the support-side score, I explain how I calculate CSAT separately.

Frequently asked questions

Is a negative NPS bad?

It means you have more detractors than promoters. It is a warning, but the trend and the comments matter more than one quarter's number.

Do passives count in the NPS calculation?

Yes, in the total. Passives are not subtracted, but they sit in the denominator, so more passives pull both percentages down and move the score toward 0.

How many responses do I need for a reliable NPS?

Around 100 per segment is my minimum before I compare segments, and even then the margin of error is about ±15 points. At 400 answers it is closer to ±8.

How often should I measure NPS?

Quarterly works for most B2B teams. Monthly is fine if each segment still gets about 100 answers.

Sources

• Measuring your Net Promoter Score, Bain & Company
• The One Number You Need to Grow, Frederick F. Reichheld, Harvard Business Review (2003)