Measuring customer satisfaction: NPS, CSAT or reviews?

Start with your goal, not the questionnaire

A customer satisfaction survey is only useful if you know what you will do with the results. Do you want to know whether customers recommend you, whether a particular touchpoint runs smoothly, or why customers drop out? Each goal calls for a different question and a different moment.

Keep it small: one main question, one open question and at most a few extra questions. Every additional question reduces the chance that someone finishes the questionnaire.

Decide in advance who will look at the results, and when. A survey that nobody reads costs customers time without changing anything.

NPS, CSAT and CES: three well-known measures

The Net Promoter Score (NPS) asks how likely someone is to recommend you, on a scale of 0 to 10. Scores of 9 and 10 count as promoters, 0 to 6 as detractors. The NPS is the percentage of promoters minus the percentage of detractors, so it ranges from -100 to +100. The measure became well known through Fred Reichheld’s 2003 article in Harvard Business Review.

CSAT (Customer Satisfaction Score) measures satisfaction with one specific experience, for example on a scale of 1 to 5. Often the share of answers scoring 4 or 5 counts as satisfied. CES (Customer Effort Score) asks how much effort it took to get something sorted, such as a return or a question to customer service.

None of these numbers is ‘the’ truth. NPS is handy for tracking over time, but researchers have questioned the claim that a single number predicts growth. So use a score as a pointer, and always read the comments. Roughly, you choose like this:

• NPS: a general picture of loyalty, for example twice a year
• CSAT: right after a delivery, appointment or customer contact
• CES: after a process that should run smoothly, such as a return

The right questions at the right time

Ask neutral questions. ‘How satisfied are you with the delivery?’ is better than ‘What did you think of our super-fast delivery?’. Always follow a score with an open question, such as ‘What is the main reason for your score?’. That explanation is often more valuable than the number.

Timing matters a lot. Ask about delivery shortly after it arrives, and about the product only once the customer has been able to use it. For service businesses, ask ideally right after the job is finished, while the experience is still fresh.

How many responses do you need?

With ten responses you cannot draw firm conclusions, but you do discover recurring themes. If you want to compare numbers between months, you need more responses. With about 100 responses, a percentage can still be around 10 percentage points higher or lower by chance; with 400 responses it is about 5 percentage points.

Also look at who responds. Very satisfied and very dissatisfied customers often respond more readily than those with a more neutral view. So compare mainly with your own earlier measurements carried out in the same way.

Increase the response rate by keeping the questionnaire short, testing it on your phone and making the invitation personal. Never tie a reward to a positive answer; that skews the results.

Privacy and fairness

In your invitation, say who you are, what you will use the answers for and how long it takes. Do not ask for more personal data than necessary and mention the survey in your privacy notice. If your email also contains offers or other advertising, the rules for marketing email apply; in that case, make sure there is at least an easy way to unsubscribe.

Do not keep answers longer than necessary, and only link them to a customer number if you really need to, for example to resolve a complaint.

A private survey is not the same as a public review. It is tempting to send only satisfied customers on to a review platform, but that is called review gating. Google prohibits selectively asking for positive reviews, and painting an overly rosy picture this way risks misleading consumers. So invite all customers to leave a review in the same way, regardless of their survey score.

A simple set-up in five steps

Want to start tomorrow? Keep it simple. Example questions for an online shop: ‘How satisfied are you with your order, on a scale of 1 to 5?’ and ‘What should we improve first?’. For a service business: ‘How likely are you to recommend us, from 0 to 10?’ and ‘What was the deciding factor for you?’.

This set-up suits most small businesses:

• Choose one goal, for example: how do customers experience our delivery?
• Ask one score question (CSAT or NPS) and one open question
• Pick a fixed moment, such as three days after delivery
• Send the questionnaire to all customers in that period, not to a selection
• Review the answers every month and pick one thing to improve

From results to improvement

Put open answers in a simple table and tag each one with one or two themes, such as delivery time, packaging or reachability. Choose one theme per quarter to improve and tell customers what you have changed. That way they see that taking part is worthwhile.

Public reviews complement this: they are less structured but visible to new customers. With the AI review analysis in your trustmark.site account, you can discover recurring themes in reviews. The result remains an aid and does not replace your own judgement.

• Analysing reviews with AI
• Collecting reviews: the complete guide
• See the review modules