How to Calculate Your NPS Score (Formula, Examples, and Common Mistakes)

How do you calculate an NPS score?#

To calculate your NPS score, subtract the percentage of respondents who are detractors (people who answered 0–6 on the "how likely are you to recommend us" question) from the percentage who are promoters (people who answered 9 or 10); the result is a single number between −100 and +100. Passives — those who answer 7 or 8 — are counted in your total response base but are left out of the subtraction itself. That one line is the entire NPS calculation, which is exactly why the NPS formula spread so fast: anyone can run it in a spreadsheet.

The formula, written plainly:

NPS = % Promoters − % Detractors

The hard part of NPS was never the arithmetic. It's knowing what to do with the number once you have it — and this guide covers both, starting with the mechanics and ending with the follow-up that actually moves the score. If you want the concept-level background first, see our explainer on what Net Promoter Score is and where it came from.

The NPS formula step by step (with a worked example)#

Calculating NPS takes four steps: collect responses on the 0–10 scale, sort each response into a segment, convert the segment counts to percentages, then subtract. Here is a worked example using 500 completed responses.

Step 1 — Ask the standard question and collect 0–10 answers. The Net Promoter question is "On a scale of 0 to 10, how likely are you to recommend [company/product] to a friend or colleague?" The 0–10 range is not arbitrary; for why it's eleven points and not a 1–5 star rating, see the NPS scale explained.

Step 2 — Sort each response into one of three segments. Every answer maps to exactly one bucket:

Step 3 — Convert each segment to a percentage of total responses. Divide each segment's count by the total number of responses. In our example of 500 responses:

Step 4 — Subtract detractors from promoters. Take the promoter percentage and subtract the detractor percentage:

NPS = 60% − 20% = +40

The passives (20%) never enter the subtraction, but they still shape the result: because they sit in your base, a wave with lots of passives will have a lower NPS than one where those same people became promoters. Once you have your number, the next question is whether +40 is any good — which depends entirely on your industry, and is covered in our breakdown of what counts as a good NPS score by industry.

Common mistakes when calculating NPS#

Most NPS errors come from mishandling the segments, the sample, or the scale — not the subtraction. These are the five that most often produce a misleading number.

• Counting passives in the math. Passives (7–8) belong in your response base but never in the promoter or detractor percentages. Folding them in — a frequent spreadsheet slip — deflates the score and breaks comparability with every other NPS on record.
• Using a 1–10 or 1–5 scale. The Net Promoter question is defined on a 0–10 scale. Shifting to 1–10 or a 5-point star rating changes where the promoter/detractor lines fall and makes your number incomparable to benchmarks.
• Calculating on too small a sample. A minimum of roughly 100 responses is needed for a stable score; below that, a single answer can swing NPS by several points. For segmenting by product or persona, aim for 200–300 responses per cell before drawing conclusions.
• Comparing waves of different sizes without noting the variance. A jump from +38 to +44 can be pure sampling noise if the two waves had very different response counts. Track the base alongside the score.
• Treating the score as the finding. The number tells you the ratio of fans to critics; it does not tell you why anyone answered the way they did. Ignoring the open-ended verbatim is the most expensive mistake of all — more on that below. Broader survey design pitfalls are covered in our guide to the NPS survey.

How often should you measure NPS?#

Measure NPS on a cadence that matches the type of NPS you're running: relational NPS is typically measured quarterly or twice a year, while transactional NPS fires right after a specific interaction. Relational NPS asks about the overall relationship and is best on a steady quarterly rhythm so you can compare like-for-like waves over time. Transactional NPS asks about a single moment — a support ticket, an onboarding step, a renewal — and is sent within hours of that event, which also explains why transactional surveys post noticeably higher response rates than relational ones. Whichever cadence you choose, hold it constant: an inconsistent schedule reintroduces the wave-comparison problem from the mistakes above, and it muddies the link between NPS movements and the customer retention signals you're ultimately trying to protect. NPS is also rarely read alone — most teams track it alongside CSAT and the other core CX metrics.

From score to action: the follow-up that matters#

The calculation is trivial; converting the score into a change in customer behavior is the genuinely hard part, and it lives entirely in the reason behind each rating. A +40 and a +40 can describe two completely different businesses — one full of price-sensitive detractors, one full of customers frustrated by a single broken workflow. The number is identical; the fix is not. This is the structural limit of the metric that our post on why traditional NPS surveys are not enough unpacks in depth, and it's why some product teams are sunsetting NPS as a standalone KPI.

The standard fix is to bolt an open-ended "why did you give that score?" box onto the survey. It helps, but a static text field collects a sentence or two from the minority who bother to type — thin input that starves any downstream customer feedback analysis. The higher-leverage move is to replace the comment box with a short conversation. Perspective AI runs AI-moderated interviews that ask the NPS question and then follow up in the customer's own words — probing a detractor on what specifically went wrong, or asking a promoter what they'd tell a colleague. This is the conversational method for capturing the why behind the score, and it's part of a broader shift from survey-based CX measurement to conversational voice-of-customer. If you've already decided the text box isn't enough, the best NPS alternatives for 2026 lays out the options.

Frequently Asked Questions#

What is the formula for NPS?#

The NPS formula is the percentage of promoters minus the percentage of detractors: NPS = % Promoters − % Detractors. Promoters answer 9 or 10 on the 0–10 recommendation question, detractors answer 0–6, and passives (7–8) are excluded from the subtraction but still count toward your total response base. The result is a whole number from −100 to +100 and is expressed as a number, not a percentage.

Do you include passives when calculating NPS?#

No — passives are excluded from the NPS subtraction, but they are not deleted from your data. Respondents who score 7 or 8 stay in your total response base, so they affect the denominator when you convert promoters and detractors to percentages. A high share of passives lowers your score indirectly by diluting the promoter and detractor percentages, even though their scores never appear in the formula itself.

Is an NPS of 40 good?#

An NPS of +40 is generally considered good, but "good" is entirely relative to your industry. According to the Net Promoter framework developed at Bain & Company, scores above 0 mean you have more promoters than detractors, and above +50 is often called excellent. Because benchmarks vary widely by sector, compare your number against industry-specific ranges rather than a universal threshold.

How many responses do you need for a reliable NPS?#

You need at least about 100 responses for a stable overall NPS, and 200–300 per segment if you plan to break the score down by product, region, or persona. Below roughly 100 responses, a single answer can move the score by several points, making wave-over-wave comparisons unreliable. Always report the response count alongside the score so readers can judge whether a change is real or sampling noise.

What is the difference between transactional and relational NPS?#

Transactional NPS measures a single interaction and is sent right after it, while relational NPS measures the overall relationship on a recurring cadence. Transactional surveys — after a support ticket or onboarding step — capture fresh, specific feedback and tend to earn higher response rates. Relational surveys, usually run quarterly or twice a year, track long-term loyalty. Many teams run both and keep each on a consistent schedule so waves stay comparable.

Conclusion#

Learning how to calculate your NPS score takes about a minute: sort responses into promoters, passives, and detractors, turn the promoter and detractor counts into percentages, and subtract. The NPS formula — introduced by Fred Reichheld in his 2003 Harvard Business Review article and refined by Bain & Company since — has stayed popular precisely because the math is so simple. But a score with no reason attached is a thermometer, not a diagnosis. The teams that actually move their number are the ones who capture why customers answered the way they did, at scale, in the customer's own words. That's where a conversation beats a comment box — see how Perspective AI turns the NPS question into a follow-up interview, or start a study and put your next NPS wave to work.