How to Run an NPS Campaign: 9 Steps

An NPS campaign is a repeatable process for asking one question, scoring the answers, and acting on them. The question is: how likely is it that you would recommend our company, product, or service to a friend or colleague? Customers answer on a 0 to 10 scale. Promoters score 9 or 10, passives score 7 or 8, and detractors score 0 through 6. The Net Promoter Score is the percentage of promoters minus the percentage of detractors, written as a whole number between -100 and 100.

The metric is easy. The campaign is where most of the value and most of the failure sit: choosing the right moment to ask, asking the right people, and doing something with the answers. This guide covers the calculation, the nine steps of a campaign, what to do with each response group, and the mistakes that make NPS data useless.

What is the Net Promoter Score?

NPS was designed as an alternative to long satisfaction questionnaires. Instead of measuring how pleased someone was with a specific interaction, it measures whether they will put their own reputation behind a recommendation. That is a higher bar, and it is why the scale is deliberately harsh: a 7 or an 8, which reads as a good grade in most contexts, counts for nothing.

Two details are worth getting right, because they are the two most commonly reported wrong. The scale starts at 0, not 1. And the output is a score, not a percentage. An NPS of 40 is "forty", not "forty percent".

How to calculate NPS

Worked example

A store collects 500 responses to the recommendation question:

• 300 respondents scored 9 or 10 → promoters = 300 / 500 = 60%
• 100 respondents scored 7 or 8 → passives = 100 / 500 = 20%
• 100 respondents scored 0 to 6 → detractors = 100 / 500 = 20%

The three groups account for every response, so the percentages sum to 100. The score is 60 − 20 = 40.

Notice what the passives do. They do not enter the formula, but they occupy 20 percent of the base. Converting half of them into promoters would lift the score from 40 to 50 without resolving a single complaint. Converting detractors does the same work twice over, since each one moves out of the subtracted group and into the added group.

Read the score in segments, not in aggregate

A single company-wide score is a headline, not a diagnosis. Split it by product line, acquisition channel, market, and customer lifecycle stage, and the number starts to point at causes. It is common for one channel or one product to be responsible for most of the detractors while the blended score looks acceptable. The same split also makes trend analysis meaningful: a flat overall score can hide a rising score in one segment and a collapsing one in another.

Two variants are worth tracking separately. Relationship NPS measures overall loyalty to the brand and is sent on a fixed cadence. Transactional NPS measures a specific event, such as a delivery or a support ticket, and is sent right after it. They answer different questions and should not be averaged together.

What to look for in an NPS survey tool

The features that matter in practice:

• Customization. Templates you can adapt, and control over colors, fonts, and layout so the survey looks like it came from you.
• Multi-channel distribution. Email, SMS, in-app, on-site, and social. If you sell in person, offline collection through a kiosk or tablet matters too.
• Survey logic. Conditional follow-up questions based on the score, so promoters and detractors are asked different things.
• Integrations. CRM, help desk, and marketing automation, so a response can trigger a ticket or an email without manual work.
• Segmented reporting. Response rate tracking, breakdowns by customer attribute, and text analysis of the open-ended answers.
• Targeting and timing controls. The ability to decide who sees the survey and when, which is what protects your response rate.

Surveys in Omniconvert Explore cover the on-site and post-purchase side of this: targeting by behavior or traffic source, triggering by event, and testing the wording and placement of the survey itself against a control.

How to run an NPS campaign in 9 steps

1. Select the survey tool

   Choose against the criteria above, and weight integration with your existing stack heavily. A tool that cannot write a response back to the customer record forces every follow-up to be manual, which is where most campaigns quietly stop.
2. Set one clear objective

   Decide what the campaign is for before you write a question. Measuring overall loyalty, evaluating a specific product, and assessing support performance are three different campaigns with different audiences and different timing. Pick one, and state what decision the result will inform.
3. Segment the audience

   Not everyone should get every survey. Segment by purchase history, lifecycle stage, or recent interaction so the question is relevant to the person answering. Targeted sends raise response rates and produce data you can act on. Customer segmentation models are a useful starting point, and Nexus by Omniconvert can build these segments from purchase behavior.
4. Write the survey: two questions, no more

   Question one is the standard scale: "On a scale of 0 to 10, how likely are you to recommend our product or service to a friend or colleague?" Question two is open-ended: "What is the main reason for your score?" Keep the wording neutral, keep the brand styling consistent, and offer the survey in the languages your customers read. Test it before it goes out.
5. Send it at the right moment

   Transactional surveys go out shortly after the event they measure, while the experience is fresh. Relationship surveys go out on a fixed cadence, typically quarterly, biannually, or annually. Match the channel to the audience, use triggers in your CRM or help desk to automate the send, and cap how often any one customer is asked. Customer data is what makes this targeting possible.
6. Set up real-time alerts

   Alert on detractor scores so support can respond the same day, on promoter scores so marketing can ask for a review or a referral, and on keywords such as "cancel" or "refund" in the open text. Route each alert to a named owner rather than a shared inbox, and define an escalation path for serious issues. Alerts are also how you improve satisfaction between measurement cycles rather than after them.
7. Calculate the score and segment it

   Group the responses, apply the formula, and then immediately break the result down by segment, product line, and time period. Track relationship and transactional NPS separately. Compare against your own history first and against industry benchmarks second.
8. Close the loop

   Reply to detractors personally, resolve the specific problem, and follow up to confirm it is fixed. Thank promoters and give them an easy way to refer someone. Then tell the wider customer base what changed because of the feedback. Closing the loop is what converts a measurement exercise into retained revenue.
9. Repeat on a fixed cadence and reassess

   Run the survey on a schedule so the trend is comparable period to period, and keep the wording and timing stable while you do. Review the campaign itself once or twice a year: response rates, whether the segments still make sense, and whether the follow-up actions actually moved the score. Share the results across product, support, and marketing, because none of those teams can act on data they never see.

What to do with promoters, passives, and detractors

The follow-up on detractors is worth staffing properly. It is the only part of the campaign where a single reply can retain a customer who had already decided to leave, and the comment usually names the problem precisely enough to act on without further research. Once the themes are stable, feed them into the wider survey program and into your retention segments in Nexus by Omniconvert, so the customers most at risk get attention before the next survey cycle.

Common NPS campaign mistakes

• No objective. A survey sent because NPS is standard practice generates data nobody owns.
• Bad timing. Asking during an unresolved support issue measures the issue, not the relationship. Asking months after a purchase measures memory.
• Leading wording. Changing the question, adding qualifiers, or preceding it with positive messaging biases the score and breaks comparability with your own history.
• No segmentation. A single blended number hides the segments that need attention.
• Ignoring passives. They are the cheapest group to convert and the easiest to lose, and they are invisible in the formula.
• No action. Collecting feedback and not responding to it is worse than not asking, because the customer now knows you asked.

One more, less obvious: comparing your score to a competitor's published score. Scores are affected by the scale wording, the channel, the timing, and the market, and two companies measuring differently are not comparable. Treat published industry benchmarks as orientation and your own trend as the real measure.

Frequently asked questions about NPS campaigns

Elena Dobre

Content Writer

Elena Dobre is a content writer with a passion for writing and online marketing. She writes about conversion rate optimization, lead generation, customer value optimization, and eCommerce growth.