Form questions

NPS survey questions

Ask two questions, three at most. The score itself on the eleven point scale, one open follow up chosen to match the score that was just given, and permission to reply. Everything else here is optional, and adding more than five questions to an NPS send is the fastest way to lose the response rate that makes the score trendable.

An NPS survey is one question plus whatever you ask next, and the second part is where nearly all the value sits. These questions give you three different follow ups so a promoter, a passive and a detractor are each asked something they can actually answer.

The questions

The score

  1. How likely are you to recommend us to a friend or colleague?Single choiceRequired

    Eleven options, 0 to 10, laid out as single choice. There is no slider or opinion scale component, so the scale is written out as options.

  2. How likely are you to recommend us to someone in your own industry?Single choice

    A narrower version for business customers, where a general recommendation question is answered about the wrong audience.

  3. How likely are you to buy from us again?Single choice

    0 to 10. Repurchase intent moves separately from recommendation and is worth tracking beside it.

  4. Which product or plan are you answering about?Dropdown

    Without this, one score covers three products and nobody can act on it.

The follow up

  1. What is the main reason for your score?Long text

    The general form. Use it when you cannot branch the follow up by score.

  2. What would it take to move your score up by two points?Long text

    The follow up for passives, who have nothing wrong to report and so answer a reason question with a shrug.

  3. What went wrong for you?Long text

    For low scores. Asking a detractor for the reason for their score invites a summary; asking what went wrong invites the incident.

  4. What do you tell people about us when it comes up?Long text

    For high scores. Far more useful than asking a promoter why they scored nine, which produces the word good.

  5. Who else at your company would answer this differently?Short text

  6. What nearly stopped you choosing us?Long text

Reason codes

  1. Which of these best describes why?Single choice

    Price, product quality, ease of use, support, reliability, something else. Coded reasons make the score reportable without anybody reading every comment.

  2. Which parts have worked well for you?Multiple choice

  3. Which parts have let you down?Multiple choice

  4. Was there one event that shaped your answer?Single choice

  5. When did that happen?Date

  6. Did you contact support about it?Single choice

Permission and context

  1. May a person from the team reply to you about this?Single choiceRequired

    Ask before collecting an address rather than after, so a no means the address is never taken.

  2. Your email, if you said yesEmail

    The email component works on a conversational form; on a classic form it renders nothing, so a classic version uses short text.

  3. How long have you been a customer?Dropdown

  4. How often do you use it?Dropdown

  5. What is your role?Short text

  6. Would you be willing to talk to us for fifteen minutes?Single choice

    Worth putting last. A yes here is the most valuable answer in the whole survey and the least likely to survive being asked first.

Which questions to make required

Two questions are required: the score, because a response without it is not an NPS response at all, and the permission question, because the answer decides whether an email address is collected. The open follow up stays optional on purpose. Requiring it costs you the scores of people who have nothing to add, and those scores count exactly as much as the ones with a paragraph attached.

The form built from these questions

A static copy of the finished form, from its own example page.

14

How likely are you to recommend Ledgerly to a colleague? *

0 means not at all likely, 10 means extremely likely.

Enter

Three ways to change this bank

For a business to business account, add the role question and the one about who else would answer differently, then send to three people per account rather than one. A single champion answering for a company is the most common way a business score ends up wrong. For a shorter version, run two questions: the score and one open follow up, with the follow up wording chosen by which band the score fell into. That is the version most software sends in app, and it is finished in under twenty seconds. For a follow up wave, write only to detractors, four to six weeks after the original send, and ask whether the thing they named has changed. Do not ask for the score again in that wave. A second score from a self selected group of unhappy customers looks like a movement in the metric and is not one.

Questions people ask

How many questions can an NPS survey carry?

Two or three. The score is the whole instrument, and everything else is a rider. Once a survey reaches five or six questions it stops being an NPS send and starts being a relationship survey, which needs a different cadence and a different sample. Keep the extras for a once a year study.

Should the follow up change with the score?

Yes, and it is the single biggest improvement you can make. A detractor asked what went wrong writes an incident. The same person asked the reason for their score writes one line. A promoter asked what they tell people gives you sales copy, while the same person asked why they scored nine writes the word good.

Is the survey anonymous?

Usually not, because most NPS sends go to a known customer and the score is attached to their account. Say so on the first screen. If you want anonymity, drop the permission and email questions, publish it as a public link with no login, and accept that you can no longer segment the result by plan or tenure.

How often should NPS go out?

Quarterly for a relationship survey, or continuously with each customer sampled no more than twice a year. Sending after every interaction turns it into a transactional survey measuring the last agent rather than the relationship. Whichever you pick, never change the wording of the score question between waves.

What response rate should I expect?

In app prompts placed after a completed task tend to run well ahead of emailed ones, and emailed NPS sends to an existing customer list commonly land in the low double digits. The number matters less than its stability: a rate that halves between waves usually means the audience changed, not the sentiment.

Can I see who gave a low score?

Only if you collected something identifying. On a signed link the response carries whatever contact field you asked for, and the export puts one column per field with the submission time first. If you plan to close the loop with detractors, ask permission before the email field so the follow up is something they agreed to.

Build this form in about a minute

The button opens the form generator with this form already described. Change the wording to match your own, generate, then edit any field.

Create a form with OneCraft

Other question banks

On writing the questions themselves, read How to write survey questions. For everything the builder can do, see the form maker. Step by step: Create a conversational form with AI, then View, export and manage your responses.

Sources

Written and checked by the OneCraft team. Last checked .