Table of Contents
- How to calculate NPS (the whole formula)
- NPS survey calculator
- Inputs
- Steps
- Second worked example (made-up counts so you can check your sheet)
- Why the number alone is useless
- 1. Bucket mix
- 2. Trend
- 3. Who specifically said what
- Mistakes that break the calculation
- What to do after you have the number
- A note on sample size (without fake benchmarks)
- When a calculator is the wrong tool
- Doing this with Testimonial
- Wrapping up
Image URL
Do not index
Do not index
AI summary
NPS is calculated as % promoters (minus) % detractors, with promoters scoring 9‑10, passives 7‑8, and detractors 0‑6; the single score hides crucial details like bucket mix, trends, and individual feedback, so always show the three percentages, sample size, and next‑step actions for each bucket, and use the provided calculator and asset guidelines to capture and act on the full insight.
Title
How to Calculate NPS (and Why the Number Alone Is Useless)
Date
Description
How to calculate Net Promoter Score: the formula, a usable NPS survey calculator, and why the number alone hides bucket mix, trend, and who said what.
Status
In progress
Current Column
In the queue
Person
Writer
You have a stack of 0–10 scores and a slide that says "+47."
Someone asks how you got there.
You open a spreadsheet, freeze, and Google the formula again.
Here it is — then why the number is the least useful part of what you just measured.
If you want the programme that turns scores into next steps, start with our guide to NPS surveys. This page is the calculator and the honesty check that sits under it.
How to calculate NPS (the whole formula)
Ask one question: How likely are you to recommend us to a friend or colleague? Answers are 0 to 10.
Bucket every response the way Qualtrics's public NPS guide states it as of October 2026 (I am using their wording, not inventing a house variant):
- Promoters: 9 or 10
- Passives: 7 or 8
- Detractors: 0 to 6
Then:
NPS = (% promoters) − (% detractors)
The result sits between −100 and +100. Passives sit in the denominator, then drop out of the subtraction. That is why a pile of 8s can look calm on a slide and still mean nobody would actually recommend you.
Worked example (Qualtrics's numbers, not ours): 70% promoters, 20% passives, 10% detractors → 70 − 10 = 60.
That is all the arithmetic. Everything below is what the single number hides.
NPS survey calculator
Use this on any export. No new URL — the calculator lives here so "how to calculate nps" and "nps survey calculator" resolve to one page.
Inputs
Count every completed response. Leave blanks at zero.
Input | Your count |
Promoters (9–10) | ___ |
Passives (7–8) | ___ |
Detractors (0–6) | ___ |
Total responses | promoters + passives + detractors = ___ |
Steps
- % promoters = (promoters ÷ total) × 100
- % detractors = (detractors ÷ total) × 100
- NPS = % promoters − % detractors
- Round however your team already reports (nearest whole number is fine). Do not invent a second formula for board slides.
Second worked example (made-up counts so you can check your sheet)
- 40 promoters, 30 passives, 30 detractors → total 100
- % promoters = 40 · % detractors = 30 · NPS = 10
Same total, different mix — 55 promoters, 10 passives, 35 detractors → NPS = 20. The slide improved. The detractor pile grew. If you only report the score, you miss that.
ASSET NEEDED — interactive calculator
What: Lightweight on-page NPS calculator (three integer inputs: promoters, passives, detractors; live % promoters, % detractors, NPS output). No separate URL/slug. Target keyword framing: "nps survey calculator".
Why here: Readers arrive for a calculator; the table above is usable offline, but an interactive widget matches the Ahrefs "nps survey calculator" demand without creating /resources/nps-calculator.
Naming: nps-survey-calculator-embed (Feather/custom block or Design Bot HTML embed)
Why the number alone is useless
A score is a temperature. It does not say who is hot, who is cold, or what to do Tuesday.
Three things the single integer erases.
1. Bucket mix
+40 can mean 50% promoters and 10% detractors — or 70% promoters and 30% detractors with almost no passives. Same number. Different company.
If you do not show the three percentages next to the score, you are asking leadership to manage a ghost.
ASSET NEEDED — diagram
What: Side-by-side bar or stacked comparison of two response sets that share the same NPS (+40) but different promoter/passive/detractor mixes; label each bar with the three percentages and the shared NPS.
Why here: The reader just heard that mix can hide under one score; the diagram makes the contradiction visible in one glance.
Naming: nps-score-breakdown-light.png
2. Trend
A one-shot +47 with no prior period is a postcard, not a diagnosis. Up five points after you shipped a fix is a signal. Flat for three quarters while you celebrate "good NPS" is a shrug with a logo.
I am not going to invent a benchmark for "good." Industry league tables are where posts quietly fabricate. Watch your series and the mix inside it.
3. Who specifically said what
The export has names. The score does not.
A 10 from a customer who just finished onboarding is a mood. A 10 from someone who renewed is a candidate for a quote. A 3 with a verbatim about billing is an owner and a deadline, not a row in a CX deck.
If nobody owns the next email per bucket, you calculated a poll. You did not run a programme. The routes — ask promoters for proof, ask passives what would make a 9, put a human on detractors — are in our guide to NPS surveys.
Mistakes that break the calculation
Counting passives as promoters. An 8 is not a soft 9. Inflating the score teaches the board the wrong lesson.
Dropping passives from the total. Percentages are of all respondents. Passives count in the denominator even though they cancel out of the subtraction.
Averaging the 0–10 scores and calling it NPS. Mean rating is a different metric. NPS is the promoter-minus-detractor percentage math above — not "add the scores and divide."
Surveying people who have not used the product. You can still compute a number. It will not mean loyalty.
Reporting NPS with n = 8. The formula still runs. The confidence does not. Say the sample size out loud or do not put the number on a slide.
Hiding the mix. If the slide shows only +47, ask for the three percentages before anyone debates "is that good."
What to do after you have the number
- Write the three percentages next to the score. Every time.
- Compare to the last period you trust — same question, similar audience.
- Open the export. Assign an owner to every detractor and a next step to every promoter you would actually put on the site.
- Do not start a second survey until those next steps shipped. Another wave of scores on top of an unread CSV is noise.
You do not need software for the arithmetic. A spreadsheet is enough. Software earns its place when the sheet is a week behind and nobody remembers who owns the 2s.
A note on sample size (without fake benchmarks)
I will not tell you "you need 100 responses" or "±3% at 95% confidence" as if this were a lab paper. Those lines usually arrive without a source, or with a source that assumed a different survey.
What I will say:
- Publish the n next to the score. Always.
- Prefer a smaller, recent, on-product sample over a large, old, mixed one.
- If leadership wants a "good NPS" debate and you only have twelve answers, the honest move is to delay the debate, not to invent precision.
The formula does not care. Your decision quality does.
When a calculator is the wrong tool
Skip the spreadsheet theatre when:
You have not decided who owns a 2. Computing −20 faster does not help.
You are about to survey people who have never used the product. You will get a number. It will describe onboarding vibes, not loyalty.
You already know the mix and nobody acted on last month's export. Another calculation is procrastination with arithmetic.
You wanted a 12-question product survey. That is not NPS. Do not bolt eleven extra questions onto the recommend score and still call the result NPS — you changed the instrument.
In those cases, fix the programme first. Our guide to NPS surveys is the page for that. Come back here when you have an export and need the math clean.
Doing this with Testimonial
NPS with auto-routing still uses the same 0–10 question and the same textbook buckets by default. You can change the thresholds if your buyers talk differently — promoter can mean "10 only" if that is how you sell. The product computes the live score, colour-codes health, and shows bucket mix and trend so you are not rebuilding the calculator every Monday.
What it adds after the number: rules per bucket (promoter → testimonial or case-study invite with identity pre-filled; passive → follow-up email; detractor → support alert). That is the programme. The formula on this page does not change.
NPS is on Ultimate +. As of 1 October 2026 that is $95/month/space, or $1,140 a year. Ultimate and Ultimate + are priced per space. The plans and pricing page is the live list. Paid plans include a 7-day free trial.
Where it breaks, including for us. A dashboard score is not a submitted testimonial. Routing still needs a human to open the alert. We do not publish a conversion rate from a 9 to a quote, because I will not invent one. And nothing in the tool replaces permission to put someone's words on your site.
If you have thirty responses and you will actually send the three follow-ups by hand, send them. You do not need us for the calculator.
Wrapping up
NPS = % promoters − % detractors. Promoters are 9–10, passives 7–8, detractors 0–6. Put the mix and the names next to the score, or the number is decoration.
If you only do one thing this week: recalculate last quarter with the table above, write the three percentages on the slide, and pick five promoters who still have no ask.
Stuck on whether your sheet is averaging scores instead of running real NPS? Send the columns — I will tell you which formula you actually used.
Written by
