Reading customer feedback beyond the NPS score

You have built a careful survey. The questions are well worded, the scales are tested, and you have added a box for free-text comments. The response rate is healthy. On paper you are golden — you can structure and analyse every answer in bulk.

But there is a catch. The single number at the heart of the survey, your Net Promoter Score, may not mean what you think it means. Interpreting customer feedback beyond the NPS score is trickier than it first appears.

People read the 0–10 scale very differently

The classic NPS questions look simple:

Would you recommend our service?

How likely are you to recommend our service?

Customers pick a number, and you learn what they think of you. Straightforward — except that people do not perceive the 0–10 scale in the same way.

By analysing free-text comments alongside the score people gave, you can test how well the number actually reflects their opinion. A reasonable starting assumption is that people who write positive comments are promoters, and those who write negative comments are detractors who may speak poorly of you to others. The data does not always cooperate with that assumption.

When the score and the comment disagree

Comparing the calculated NPS (the share giving 9–10 minus the share giving 0–6) with a sentiment-based score (the share of positive comments minus the share of negative comments) reveals a gap.

• In some sectors, the NPS from numerical scores is noticeably lower than the NPS implied by the text. People are consistently more conservative when scoring than when they explain themselves in words.

• In other sectors the two match closely.

The point is not that one method is right and the other wrong. It is that looking only at numbers can give you an incomplete — and sometimes actively misleading — picture.

Why the text is where the value hides

Analysing the text your customers write tells you far more than a number can. Most valuably, it surfaces the customers who give a high score but write a negative comment. These are the people on the verge of quietly drifting away — or worse, switching to a competitor — while your dashboard still counts them as happy.

Text analysis lets you:

• Catch mismatches between score and sentiment before they turn into churn.
• Understand the "why" behind a rating, not just the rating itself.
• Segment feedback by topic, so you know which issues drive detractors.
• Compare fairly across markets and cultures, where the same score can carry very different meaning.

Turning feedback into action

A score tells you where you stand. The comments tell you what to do about it. When you combine automatic sentiment and topic analysis with your survey data, feedback stops being a scoreboard and becomes a to-do list: fix the recurring complaint, reassure the at-risk promoters, and double down on what genuine advocates praise.

Frequently asked questions

What is NPS?

Net Promoter Score asks customers how likely they are to recommend you on a 0–10 scale, then subtracts the share of detractors (0–6) from the share of promoters (9–10). It is a simple, widely used loyalty metric.

Why is the NPS score alone not enough?

Because people interpret the scale differently and the number carries no explanation. Two customers giving the same score can mean completely different things, which you only discover by reading their comments.

How does text analysis improve NPS?

It classifies the sentiment and topics in free-text comments across every response, so you can see whether the words match the score and understand the reasons behind each rating.

Which customers should I worry about most?

Often the ones who give a high score but leave a negative comment. They look loyal in the numbers but are the most likely to leave.

Where to go next

• The pillar guide: What is conversation analytics?
• Score every interaction: Automatic quality scoring
• Deeper on scores: NPS analysis: looking behind the score

Want to see what your customers really mean, not just how they scored you? Book a demo and we will analyse your feedback text alongside the numbers.