NPS Survey Questions: 21 Examples You Can Copy
- The NPS question is fixed: On a scale of 0 to 10, how likely are you to recommend us to a friend or colleague? Rewriting the scale or the wording makes your score uncomparable.
- Scores of 9 to 10 are promoters, 7 to 8 are passives and 0 to 6 are detractors. NPS is the promoter percentage minus the detractor percentage, on a range of -100 to +100.
- One rating question plus one open-ended follow-up is the minimum useful survey. Every question after that trades response rate for detail.
- Branch your follow-up by score. The question that helps a detractor explain a failure is the wrong question to ask a promoter.
- Send NPS twice: right after checkout for the buying experience, and a few days after delivery for the product. The gap between the two scores tells you where the problem is.
The Net Promoter Score survey has one fixed question — On a scale of 0 to 10, how likely are you to recommend us to a friend or colleague? — and everything useful happens in the question you ask after it. The rating gives you a number. The follow-up gives you the reason, and the reason is the part you can actually fix.
Below are 21 NPS question examples you can copy, grouped by who you are asking rather than numbered in a list you have to read end to end. Use the ones that match the group that answered you: detractors need to explain a failure, passives need to tell you what was missing, promoters need to tell you what worked. The article also covers the three question types, where each one earns its place, when to send the survey, and the wording mistakes that quietly wreck your answers.
What is the standard NPS question and how does scoring work?
The question and the 0 to 10 scale come from Fred Reichheld's December 2003 Harvard Business Review article, The One Number You Need to Grow, written while he was a fellow at Bain & Company. That is the whole reason NPS is comparable across companies: everybody asks the same thing the same way. The moment you change it to a 1 to 10 scale, or to "how satisfied are you", you have a perfectly good survey question and a score that means nothing outside your own spreadsheet.
Worked example. You collect 1,000 responses: 400 promoters (40%), 300 passives (30%) and 300 detractors (30%). Your NPS is 40 − 30 = +10. Positive, but with three detractors for every four promoters there is a lot to fix — and the score alone will never tell you what. That is the job of the follow-up question.
If you need the full method rather than the questions, our Ultimate NPS Guide covers measurement and analysis, and the NPS definition page explains why a single average score hides more than it reveals. For where your number sits against your category, see the NPS benchmarks.
What are 21 NPS survey question examples by audience?
Copy these as they are, or swap in your product and brand name where the brackets are. They are grouped by job, so you can pick the smallest set that answers the question you actually have.
Ask everyone, straight after the score
If your tool cannot branch by score, use one of these two. They are neutral enough to work for a 2 and for a 10.
- What is the main reason behind your score?
- How would you describe your shopping experience with us?
Ask detractors (0-6): find out what failed
Detractors have just told you they would warn people away. You will probably not get a second chance with them, so use this one contact to learn as much as you can. Keep the question specific — vague questions get vague answers, and "we could do better" is not something you can act on.
- What did you like least about your shopping experience?
- What specific problems did you have with [product]?
- Why would you not recommend us?
- What was missing or disappointing in your experience with us?
- What was your biggest difficulty when buying from our store?
- Can you name three things we could improve about our shop?
Ask passives (7-8): find out what would have made it a 9
Passives are the most under-used group in NPS. They do not count towards your score, they rarely complain, and they are one competitor promotion away from leaving. They are also the group with the shortest distance to travel, so ask them what that distance is made of.
- Is there any way we can improve your experience next time?
- What can we do to serve you better?
- What is the one thing that would have made this a better experience?
- What would you need for [product] to become essential to you?
- What would make you buy from a competitor instead?
Ask promoters (9-10): capture the words that sell
Promoters are free copywriting. They will tell you, in their own vocabulary, exactly which benefit convinced them — and that phrasing is worth more on a product page than anything written internally.
- What convinced you to become a customer?
- What are the main benefits of using [product]?
- Which features do you value most in [product]?
- What has [product] changed for you?
- How does [product] help you?
Ask when you are researching a specific decision
These are for the times you are not running NPS as a health check but using it to settle a question — a roadmap decision, a repositioning, a category you are unsure about. Add one of them, not all three.
- What problem were you hoping to solve with [product]?
- How well does [product] meet your needs?
- What was missing in your experience with [brand name]?
Match the question to the score you received and you get answers you can group, count and act on. Ask the same generic follow-up of everybody and you get a folder of comments nobody reads. If you would rather start from a finished survey than assemble one, use our NPS survey template, and see promoters, passives and detractors for what to do with each group afterwards.
Ask these questions on-site, branched by score, and A/B test the fix in the same platform.
See how surveys work in Omniconvert Explore →What are the three types of NPS questions?
| Question type | What it gives you | What it costs | Use it for |
|---|---|---|---|
| Scale (0-10) | The score itself, comparable over time and against benchmarks | Almost nothing for the respondent; no explanation for you | The opening question of every NPS survey, without exception |
| Multiple choice | Countable answers you can chart the day they arrive | Only returns options you wrote, so it confirms rather than discovers | Second-round surveys, once open answers have told you what the options are |
| Open-ended | The reason behind the score, in the customer's own vocabulary | More effort to answer, more effort to read and categorise | The follow-up to the score, where the actionable insight lives |
The trap with multiple choice is that it feels efficient. You get tidy percentages and a chart by lunchtime. But a respondent can only pick from the reasons you already thought of, which means the survey can never tell you about the failure you do not know exists. Open answers can — and they hand you the customer's own phrasing, which matters, because "the size guide was wrong" and "it didn't fit" are two different problems with two different fixes.
The realistic compromise: run open-ended follow-ups first, read a few hundred answers, then convert the recurring themes into a multiple-choice question for the ongoing survey and keep one open box at the end. NPS is not the only survey worth running either — see the wider set of customer satisfaction surveys for questions outside loyalty.
How do you structure an NPS survey so people finish it?
-
Lead with the rating, unchangedThe 0 to 10 question comes first and sets the tone. Render the scale directly in the email or the on-site widget so the first click costs nothing. A survey that starts with "click here to take our survey" loses most of its respondents before the first question.
-
Branch the second question by scoreRoute 0-6 to a detractor question, 7-8 to a passive question, 9-10 to a promoter question. Same survey, three paths. This single change is what turns a comment folder into three groups you can act on separately.
-
Stop at two or three questionsEvery question after the follow-up trades response rate for detail. If you need depth, run a longer survey with a small sample rather than lengthening the one everyone receives.
-
Make the layout readableOne question per screen, generous type, no dense blocks of small text, and a progress indicator if there is more than one screen. People abandon surveys that look like work before they read a word of them.
-
Test the length instead of guessing itTwo minutes is nothing to one customer and too much to another, so no benchmark helps you here. Run a two-question version against a longer one, compare completion rates, and see where in the longer version people drop out. That drop-off point is your real maximum length.
When and where should you ask NPS questions?
Pre-delivery, right after checkout. Trigger it on the order confirmation page or in the confirmation email. At this point you are measuring one thing only: was buying from us easy? Findability, product information, delivery options, checkout friction. Nothing the customer says here is about the product, because they have not seen it yet — which is exactly what makes the answer clean.
Post-delivery, a few days after arrival. Wait long enough for the customer to have used the product, and not a day longer. For consumables that might be three days; for something like a mattress, two weeks. This is the score that tells you whether you delivered the value you promised, which is the question that decides whether they come back.
For channel, email is the default because it reaches customers after they have left the site, and it maps naturally onto the post-purchase sequence you already run. On-site widgets work better for pre-delivery moments and for people who never open marketing email. Whichever you pick, connect it to the rest of your customer lifecycle marketing so the survey lands as part of a conversation rather than an interruption, and see our roundup of NPS software if you are still choosing a tool.
Which wording mistakes ruin NPS answers?
- Double-barrelled questions. "Were you satisfied with the product and the shipping?" If the product was good and the shipping was late, no honest answer exists. Split it, or drop the half you care about less. Same problem in "Was it easy to find the products and complete checkout?"
- Leading questions. "How much did you enjoy our fast delivery?" assumes the delivery was fast and that they enjoyed it. You will get agreement and learn nothing. Ask "how was the delivery?" and let the answer be bad.
- Vague questions. "How can we improve?" is technically open-ended and practically useless. "What did you like least about your shopping experience?" points at one thing and gets one thing back.
- Internal jargon. If you ask about the "PDP" or the "onboarding flow", the customer either guesses or leaves. Use the words they would use: the product page, getting started.
There is a fifth, quieter mistake: asking a question you have no intention of acting on. If nobody in the business owns the answers to "what would make you buy from a competitor instead?", cut the question. A shorter survey that changes something beats a thorough one that files itself.
How do you turn NPS answers into action?
Read the open answers before you look at the score. The number tells you the direction; the words tell you the cause. Group them into themes — sizing, delivery times, unclear returns, missing product detail — and count each theme rather than reacting to the loudest single comment.
Then decide by value, not by volume. A complaint from twenty first-time discount buyers and a complaint from twenty repeat customers are not the same complaint. Nexus by Omniconvert segments customers by recency, frequency and monetary value, so you can weight NPS feedback by who it came from and fix what your best customers are unhappy about first.
Finally, turn the theme into a change and test it. If detractors say product information was thin, that is a hypothesis about the product page, not a fact. Omniconvert Explore runs the on-site surveys that collect this feedback and the A/B tests that validate the fix, across 70,000+ experiments on 7,000+ websites in 15+ industries, with a 23.2% average uplift for the changes that win. If you want more on the operating side of this, see our NPS survey best practices.
NPS survey questions FAQ
The standard NPS question is: On a scale of 0 to 10, how likely are you to recommend us to a friend or colleague? The 0 to 10 scale comes from Fred Reichheld's 2003 Harvard Business Review article The One Number You Need to Grow. Scores of 9 to 10 are promoters, 7 to 8 are passives and 0 to 6 are detractors. Use the wording as it is, because changing the scale or the phrasing makes your score impossible to compare with anyone else's.
The best general follow-up is an open-ended one: What is the main reason behind your score? It works for every respondent and it captures the why in the customer's own words. If your survey tool lets you branch by score, replace it with a follow-up written for that group: ask detractors what went wrong, ask passives what would have made the experience better, and ask promoters what convinced them to buy.
One question is enough to produce a score. Two is the practical minimum: the 0 to 10 rating plus one open-ended follow-up that explains it. Most brands stop at two or three, because every extra question costs response rate. If you need more depth, run a longer survey with a smaller sample rather than adding questions to the survey everyone receives.
Scale questions (the 0 to 10 rating) produce the score but not the reason. Multiple-choice questions are fast to answer and easy to analyse, but they only confirm the causes you already suspected. Open-ended questions are the only type that captures problems you have not thought of, in the customer's own vocabulary, at the cost of more effort for the respondent and more analysis for you.
Send NPS at two moments. The first is right after checkout, which measures the buying experience while it is fresh and before the product can influence the answer. The second is a few days after delivery, once the customer has actually used what they bought, which measures whether the product kept the promise. Comparing the two scores tells you whether your problem is in the shop or in the product.
A double-barrelled question asks about two things at once, for example: Were you satisfied with the product and the shipping? If the product was good and the shipping was late, no single answer is true, so the respondent picks one subject or abandons the survey. Split every such question into two, or pick the one subject you actually need to learn about.
Yes. Promoters are the cheapest source of marketing copy you have. Asking them what convinced them to buy and what they value most gives you the exact words your best customers use, which you can then test in product page copy, ads and email subject lines. Promoters are also the group to ask for reviews and referrals, because they have just told you they would recommend you.
Keep the survey short, ask one thing per question, show a progress indicator, and send it when the experience is still fresh. Put the 0 to 10 scale directly in the email or on the page so the first click costs nothing. Then test the length: run a two-question version against a longer one and compare completion, not just opens. There is no universal ideal length, so the only reliable answer is your own test.
Pick two questions, not twenty. Keep the 0 to 10 rating exactly as it is written, then add one open-ended follow-up branched by score: What did you like least about your shopping experience? for detractors, Is there any way we can improve your experience next time? for passives, and What convinced you to become a customer? for promoters. Send it right after checkout, and send a second one a few days after delivery. Give it two weeks, read the open answers before you look at the score, and note the words customers use for their problems. Those words are your next round of product page copy and your next A/B test. Only add a third question once the first two are producing answers you act on.
Ask better questions, then test the answers
Omniconvert Explore runs on-site and post-purchase surveys next to your A/B tests, so the feedback you collect turns straight into an experiment. Across 70,000+ experiments on 7,000+ websites in 15+ industries, the average uplift for tested changes is 23.2%.