What is a good NPS score? What to compare yours against
A good NPS score is less about a magic number than about comparing the same question, at the same moment, over time. Here is how to answer your manager.

Your manager asks whether our NPS is good. It is a fair question, and it is hard to answer in one line. This guide is for product managers who get asked it. If you have wondered what is a good NPS score, the honest answer depends less on the number than on how you collected it. By the end you will have a short answer for your manager, a method you can repeat, and a rule for when a change in the score is real.
In brief
- NPS runs from −100 to +100. Above 0 is a start, above 30 is usually good, above 50 is usually very good.
- Industry benchmarks vary widely and mislead, because channel, timing and who answers move the score more than the product does.
- The comparison that holds is your own score: same question, same moment, over time.
- One small change is noise. Look for a move bigger than the margin of error, repeated.
Step 1: Work out the number properly
Net Promoter Score comes from one question: how likely are you to recommend us to a friend or colleague, from 0 to 10. People who answer 9 or 10 are promoters. People who answer 7 or 8 are passives. People who answer 0 to 6 are detractors.
The score is the percentage of promoters minus the percentage of detractors. Passives count in the total but not in the subtraction. If you want to check your arithmetic, you can enter your counts in PikoPM's free NPS calculator.

Step 2: Read it on the scale
The score can sit anywhere from −100, where everyone is a detractor, to +100, where everyone is a promoter. Almost no one lives at either end. These are rules of thumb that PMs commonly use, not standards.
| Score | What it usually means |
|---|---|
| Below 0 | More people would warn others off than recommend you. Find out why before anything else. |
| 0 to 30 | More fans than critics, but a lot of room. Common for products still finding their fit. |
| Above 30 | Usually read as good. Most customers who feel strongly feel positively. |
| Above 50 | Usually read as very good. Worth asking whether your sample is skewed toward happy users. |
These bands tell you the direction. They do not tell you whether you are doing well against anyone else.
Step 3: Be wary of industry benchmarks
You will find tables of average NPS by industry. Treat them with care. Published benchmarks vary widely from one source to the next, and most do not say how the data was gathered. That matters, because three things change the score more than the product does.
Channel
An in-app pop-up, an email survey and a phone call draw different answers from the same customers.
Timing
Asking right after a successful task gets warmer answers than asking after a support ticket or a price rise.
Who answers
Active users reply more than lapsed ones. If unhappy people stay silent, the score rises without anyone being happier.
Another company may ask a different group, at a different moment, in a different way. Their number is not measuring the same thing as yours.
Step 4: Fix the method
The comparison that holds is your own score, with the same question, at the same moment, over time. So write the method down once and stop changing it.
- The exact question wording and the 0 to 10 scale.
- The moment it is asked: for example, thirty days after sign-up, or after a defined action.
- The channel: in-app, email or other. Do not mix them in one score.
- Who is invited, and who is left out.
- The period you report: monthly or quarterly.
If you must change any of these, mark the date on your chart. The score before and after is not the same series.
Step 5: Compare against yourself over time
Plot your score for each period, with the response count beside it. Look at the direction across several periods, not the gap between two. Then split it, by segment, plan or tenure. A flat overall score can hide one group rising and another falling, and that is where the useful questions are.

Step 6: Know when a change is a trend
NPS from a small sample jumps around. The more mixed your answers, the wider the swing. As a rough guide, with about 100 responses and a typical mix of promoters, passives and detractors, the score has a margin of error of around 15 points either way at 95% confidence. With about 400 responses it falls to roughly 7 or 8 points. It shrinks with the square root of the sample, so quadrupling the responses only halves the margin.
Two rules follow from this.
- Do not call a move real unless it is larger than the margin of error for your response count.
- Prefer a move that holds for two or three periods in a row over one big jump.
Worked example
Illustrative numbers, not real data
Say a team surveys 100 customers in one quarter. 50 are promoters, 30 are passives and 20 are detractors. That is 50% minus 20%, so the NPS is +30.
Next quarter, with the same question, moment and channel, 100 customers answer again. 55 are promoters, 27 are passives and 18 are detractors. That is 55% minus 18%, so the NPS is +37.
It looks like a seven-point gain. But with 100 responses each, the margin is about 15 points, so +30 could plausibly be anywhere from about +15 to +45. Seven points sits well inside that. The team should not claim an improvement. It should keep the method, collect another quarter, and watch whether the score holds above +30.
Asked whether the NPS is good, the PM can answer: it is +37, which is usually read as good. It is not clearly better than last quarter. We measure it the same way each time, so we will compare it to ourselves, not to other companies.
Step 7: Answer your manager
Keep it short. State the number and what band it sits in. Say how it is collected. Say how it has moved against your own earlier scores, and whether that move is larger than the noise. If they ask about competitors, explain that other companies' numbers come from different methods and are not safe to compare. Offer to track yours consistently instead.


