

Your NPS dropped six points. Cue the theories, the concerned faces, and a meeting called "Quick NPS sync" that will definitely not be quick. Everyone wants to know why, but the score won't tell you. Your customers will.
Net Promoter Score (NPS) measures how likely customers are to recommend your company, product, or service. Calculate it by subtracting the percentage of detractors from the percentage of promoters. The result sits between -100 and +100.
That gives you the number. To make it useful, connect the score to open-ended feedback and customer context. The practical framework has four stages: Measure, Explain, Prioritize, and Act.
NPS has survived more predicted finales than a long-running TV franchise. Back in 2021, Gartner predicted that more than 75% of organizations would abandon NPS as a customer service and support success metric by 2025. The concern was fair. A broad loyalty score can show that something changed, but it rarely tells teams what to do next.
NPS may have slipped down the priority list, but it still forms part of a wider CX measurement stack. There’s good reason for that. An analysis comparing competitors across multiple industries found that NPS explained roughly 20% to 60% of the variation in organic growth. That’s not just background noise.
NPS works best as a pulse check. It shows if customer relationships are getting stronger or weaker, while open-ended feedback reveals what’s driving the change. Together, they help CX teams move beyond collecting and reporting metrics by connecting the score to explanation, context, and action.
Used this way, NPS can become the beating heart of a Voice of the Customer program, turning customer signals into decisions. And the numbers don’t lie: Across thousands of projects created by Caplena customers, the share of new projects with "NPS" in the name remained stable over the past five years, at more than three times the share of new projects with "CSAT" in the name.
Start with the standard question. "How likely are you to recommend [company, product, or service] to a friend or colleague?" Respondents declare likelihood to recommend on a scale from 0 to 10.
|
Score |
Group |
What it indicates |
|
9–10 |
Promoters |
Highly likely to recommend |
|
7–8 |
Passives |
Positive or neutral, but less likely to advocate |
|
0–6 |
Detractors |
Unlikely to recommend |
The formula stays simple. If you work in Customer Experience, Research and Insights, you already know what follows. But here's a detailed explanation for readers who may need a refresher.
If 75% of respondents are promoters, 15% are detractors, and 10% are passives, your NPS is 60. Passives count toward the respondent total but not either side of the subtraction. If you want to track how that score changes over time, Caplena's Reports include dedicated score and trend views. Read more about it in our NPS score documentation.
Great. But the rating still needs a reason, right? Right. So pair it with one open-ended question. "What is the main reason for your score?" Use relationship NPS to track overall loyalty at regular intervals and transactional NPS after a meaningful touchpoint, such as onboarding, a purchase, or a support interaction. Keep the survey short, use neutral wording, and match its timing and cadence to the decision you want to make.
Capture market, product, customer type, touchpoint, and date so you can compare meaningful segments. Match your NPS tracking cadence to the journey, response volume, and speed of action, and avoid drawing conclusions from segments with not enough responses.
The best way to analyze NPS survey feedback is to combine the score with text comments, topics, topic-level sentiment, segments, and relevant quantitative variables. You can then move from "NPS fell" to "NPS fell among new customers in one market, and onboarding clarity appears strongly associated with the change." The score suddenly stops shrugging and actually starts explaining itself.
A Topic Collection, often called a codebook in market research, turns open-ended responses into structured evidence. Topics should overlap as little as possible while still covering the feedback that matters.
New products, phrases, markets, and languages appear over time. A human-in-the-loop workflow lets AI handle scale while researchers review assignments, refine topic descriptions, and fine-tune the analysis. Caplena's CX feedback analysis supports topic and sentiment analysis in 100+ languages while keeping the analyst in control.
Wir wussten, dass in unserem offenen Feedback wertvolle Informationen enthalten waren, aber das manuelle Analysieren nur von Teilproben bedeutete, dass wir uns nicht immer über das Gesamtbild sicher sein konnten.
Manager Market Research
Topic frequency shows what customers mention. Topic-level sentiment shows how they feel about it. "Delivery" may attract praise for speed and criticism for missed time slots. Combining those comments into one total hides the decision you need to make.
A customer can praise your staff and criticize checkout in the same text comment. Overall sentiment would flatten that story while topic-level sentiment keeps both signals visible.
Driver analysis is a statistical method that connects the topics customers mention with NPS. It shows which topics are associated with higher or lower scores and how strong those relationships are. It doesn’t prove causation, but it shows you where to investigate and prioritize. Frequency and impact are different. A common topic may have little relationship with NPS, while a smaller issue may be strongly associated with detractor scores.
Compare drivers across customer groups, markets, products, journeys, and periods. If onboarding sentiment declines among new enterprise customers before the overall score changes, you’ve found an at-risk segment and a practical question to investigate (hurray!).
Track changes in topics, sentiment, and customer segments to explain NPS movement and spot early shifts before they affect the headline score. Check the response volume and statistical significance before treating an anomaly as meaningful. Look for positive anomalies as well. When one market, team, or journey creates more promoters, study the pattern and see whether it can be repeated elsewhere.
Customers don't experience your departments as separate Marvel-style universes. Surveys, reviews, support conversations, app feedback, and purchase history all describe the same relationship from different angles.
Multiple source feedback analysis brings those signals together while preserving where each response came from. That context is important because invited surveys and public reviews have different audiences, sampling methods, and intent. Caplena's Smart Columns can align labels and create consistent variables for segmentation. For long-term tracking, test Topic Collection changes separately before applying them to production data so trends remain comparable as customer language evolves.
Once the signals are connected, package them for a decision rather than serving executives a data buffet. An executive-ready NPS update should answer four questions.
What changed?
Why did it change?
Who is affected?
What should happen next?
Show score movement, potential drivers, affected segments, representative text comments, and a proposed owner. Caplena's NPS and CSAT Reports connect scores, open text, dates, and segments. Its reporting and analytics capabilities let stakeholders explore the evidence.
Caplena has helped us move beyond looking at NPS primarily as a score and focus much more on the customer feedback behind it. By structuring qualitative feedback into clear topics and drivers, we can make customer insights more accessible and relevant across the business."
Head of Customer Strategy & Growth

Improving NPS starts with improving the experiences behind it. Use the score to decide where to act, rather than treating the number itself as the goal.
Use driver analysis to compare each topic's frequency with the strength of its relationship to NPS. Focus on potential drivers affecting valuable or at-risk segments, then validate the pattern with customer comments and operational data before deciding where to invest.
Monitor important topics, sentiment shifts, customer segments, and unusual movements in two ways. Set Alerts when you know which threshold you want to watch. For developments you didn't know to look for, Insights Radar monitors Caplena Report views and surfaces statistically significant patterns worth investigating, without prompts or predefined thresholds. Your team still decides what matters and what to do. The AI handles the watchkeeping, not the judgment.
Turn each validated insight into an owned action, then return to the data to see what changed. In Caplena, any insight surfaced by Insights Radar or spotted by a human can become a Task in one click, with an owner, priority, due date, status, and permanent link to the source data. This keeps the evidence attached as work moves from discovery to done.
Apply the same process to positive signals so successful experiences can travel across markets or teams.
For us, Caplena is an important tool for prioritizing our customer experience initiatives. Beyond identifying pain points, we can quantify how frequently customers mention specific issues and track how these topics evolve over time. This gives us a much clearer picture of where the biggest challenges are and helps us focus our resources on the improvements that matter most to our customers.”
Head of Customer Strategy & Growth

Have a specific NPS question? These quick answers cover analysis, root causes, benchmarks, and at-risk customer segments.
What should an NPS survey analysis tool do? |
| It should categorize verbatim responses, apply topic-level sentiment, connect text with scores and customer variables, find potential drivers, support multiple languages, and make evidence actionable. |
What is the best way to analyze NPS feedback? |
| Measure the score, explain it with open-ended feedback, prioritize by impact and context, then give the strongest findings an owner. |
What is a good NPS score? |
| A good NPS score depends on your industry, market, audience, channel, and method. Use a relevant NPS benchmark, but give greater weight to your consistent trend and the experiences behind it. |
How can NPS analysis identify at-risk customer segments? |
|
Compare NPS, topics, and topic-level sentiment across customer groups and periods. A deteriorating experience within one segment can reveal risk before the overall score moves. 💡 Tip: Detractors might vary regarding their level of dissatisfaction / disappointment or even anger / frustration. An LLM-based Smart Column (a special Caplena feature) allows you to simply segment this group based on the emotion and tonality in their feedback. This helps you isolate that hopefully small group of really frustrated customers, to |
NPS earns its place when it guides action. The score shows direction, open-ended feedback provides the explanation, and clear ownership turns understanding into change. Otherwise, NPS risks becoming a number with excellent attendance and very little influence.
Caplena brings those parts together in a feedback intelligence layer, from multilingual NPS analysis to shared Reports and action. Go ahead and take the Product Tour or book a call to see how the workflow fits your NPS program.