How do you calculate NPS without losing useful detail?

Sort valid responses into three groups before doing any arithmetic. A promoter gives 9 or 10, a passive gives 7 or 8, and a detractor gives anything from 0 through 6. These are the standard NPS categories. Source: Qualtrics category definitions.

The NPS formula

NPS = percentage of promoters − percentage of detractors. Calculate both percentages using all valid ratings, including passives. The result is normally reported as a score rather than a percentage. NPS methodology.

Hypothetical example: an independent repair shop receives 40 valid ratings. There are 22 promoters, 10 passives and 8 detractors. Its promoter share is 55%, its detractor share is 20%, and its NPS is +35.

Those figures illustrate the arithmetic; they are not results from a real shop. In your own report, keep the response count beside the score. Also preserve the category totals so another person can check the calculation.

Use a simple checking routine

• Confirm that every rating is between 0 and 10.
• Keep unanswered questions out of the rating count.
• Decide how to handle duplicate submissions before calculating.
• Check that the three category totals equal the valid response count.
• Do not replace the formula with the average rating.

For an initial exercise, use a spreadsheet with columns for rating, category, comment, issue owner and next action. Test your calculation with a few deliberately chosen ratings. Include the category boundaries, especially 6, 7, 8 and 9.

What should your NPS survey actually ask?

Ask about recommendation likelihood, then ask for the reason behind the rating. The standard question uses a scale from 0, meaning not at all likely, to 10, meaning extremely likely. Qualtrics survey guidance.

A short survey you can adapt

How likely are you to recommend our business to a friend or colleague? Choose a number from 0 to 10.

What is the main reason for your rating?

This is suggested wording, not a customer quotation. Replace “our business” with the name customers recognize. Show the scale labels clearly and leave the explanation as an open text field.

Resist adding every question your team wants answered. Qualtrics recommends avoiding demographic questions when they are unnecessary or the information already exists in your records. Guidance on keeping the survey focused.

Choose the relationship you want to measure

A relationship survey asks about the business overall. A transactional survey follows a particular interaction, such as a purchase or support contact; Qualtrics advises against using NPS for this narrow purpose. Relationship versus transactional NPS.

My starting recommendation is to define a customer group with enough experience to judge your business. Write down who qualifies and when you will contact them. Keep that rule available when you repeat the exercise.

For a hypothetical subscription service, you might invite established customers at a planned review point. For a project business, choose a point when customers have experienced the completed work. Treat these as design choices to test, rather than universal timing rules.

Start by checking the invitation and survey on a phone. Once the wording and follow-up process are settled, use the customer survey automation guide to plan repeatable delivery. Keep a clear rule for avoiding repeated invitations to the same person.

What should you do after a customer responds?

Read the explanation before deciding what the rating means for your next action. My recommendation is to assign each actionable comment to one person. Record what they will investigate, what they can promise and when they will reply.

A hypothetical service business example

Imagine a freelance studio receives a low rating with a comment about confusing project updates. Instead of immediately offering a discount, review the communication. Look for unclear responsibilities, missing decisions or an update the client expected but never received.

A proposed response could acknowledge the confusion and offer a clearer update schedule for the remaining work. Record whether that schedule was agreed and followed. Do not claim the intervention improved loyalty unless you have evidence.

Give every group a useful follow-up

• Detractors: confirm the problem and offer a realistic next step.
• Passives: ask which part of the experience could be easier.
• Promoters: ask what they valued and make a note to preserve it.

Do not ask customers to change their score before addressing the issue. If someone wants to recommend you, offer a simple route through your referral marketing process. Keep that invitation separate from resolving a complaint.

When comments concern several stages of buying, organize them against a customer journey map, a list of the steps customers take with your business. For example, label comments as ordering, delivery or ongoing support. Choose one stage to review rather than assigning a vague task to “improve service.”

At the next team meeting, inspect the action log. Ask whether customers received the promised response and whether the proposed change happened. Keep unresolved issues visible instead of closing them when someone merely sends an acknowledgement.

What are the limitations of NPS?

Treat NPS as one signal, with comments and customer behavior alongside it. Its ability to predict business growth better than other loyalty questions is disputed, and the score alone does not identify specific improvements. Overview of net promoter score criticism.

One score can describe different situations

Hypothetical comparison: a group with 60% promoters and 20% detractors scores +40. Another group with 40% promoters and no detractors also scores +40. The second group has a larger passive share.

That arithmetic is a reason to retain the full breakdown. In the first situation, you might prioritize understanding dissatisfaction. In the second, you might investigate what prevents more customers from giving a strong recommendation.

Context matters when comparing scores

Industry, age and location can affect NPS comparisons, according to Qualtrics. Its guide recommends looking at relevant comparisons and changes over time rather than judging a score in isolation. Interpreting NPS benchmarks.

For your first rounds, compare results collected under similar conditions. If you change the customer group, survey wording or delivery method, note the change beside the result. Avoid presenting the difference as proof that a service change worked.

Stated intention is only part of loyalty

NPS uses recommendation intentions to infer loyalty. The supplied overview also discusses combining it with customer retention information and raises concerns about overlooking passive customers. Loyalty measurement and limitations.

My recommendation is to place repeat purchases, unresolved complaints and actual referrals beside the score. For choosing a small set of measures, use the guide to business KPIs. A KPI is a key measure you use to judge progress.

Review commercial value separately through customer lifetime value, an estimate of customer value over the relationship. Do not assume the highest NPS respondent is automatically your most valuable customer. Check your own records before making spending decisions.

How can you run a useful first round this week?

Aim to finish the week with a tested survey and a clear response process. You do not need to promise a target score before establishing how you will collect and use feedback. Agree on the practical decisions the first round should inform.

1. Define the audience. Write a clear inclusion rule for customers.
2. Prepare the questions. Include recommendation likelihood and the reason.
3. Test the form. Check every rating option and the comment field.
4. Verify the calculation. Use hypothetical responses to test the category boundaries.
5. Assign follow-up. Decide who reads comments and contacts customers who agree to it.
6. Schedule a review. Reserve time to select one improvement and an owner.

At the review, separate observations from proposals. “Several comments mention unclear delivery information” is an observation only if the comments support it. “We should rewrite the dispatch message” is a proposed action to evaluate.

Before the next survey, check whether that action was completed. If it was not, resolve the implementation problem first. Keep a short record of changes so future results have useful context.