In short
- A good survey is short, sent right after the experience and has at least one open question.
- Choose the metric by the question you want answered: NPS for the relationship, CSAT for a moment, CES for ease.
- Feedback with no reply and no visible change teaches customers to stop answering.
Almost every business has sent a survey at least once. Few have got anything useful out of it. Usually not for lack of responses, but because there were too many questions, they were too vague or asked at the wrong moment - and because nobody did anything with the results.
Start with your question, not the customer’s
Before you write a single question for customers, write one for yourself: what decision do I want to make based on these answers? “I want to know whether the new check-in process works” leads to a good survey. “I want to know what customers think” leads to a 25-question one nobody finishes.
Three metrics, three different questions
| Metric | Typical question | What it’s for |
|---|---|---|
| NPS Net Promoter Score | “How likely are you to recommend us to a friend?” (0–10) | The overall relationship and loyalty. Good for trends over time. |
| CSAT Customer Satisfaction | “How satisfied were you with…?” (1–5) | A specific moment: the welcome, the service, payment. |
| CES Customer Effort Score | “How easy was it to…?” (1–7) | Friction: booking, solving a problem, paying. |
Customer Effort Score comes from research published in Harvard Business Review (Dixon, Freeman & Toman, 2010) covering more than 75,000 customers: of those reporting low effort, 94% said they intended to buy again and 88% said they would spend more. The authors’ conclusion: customers don’t necessarily want to be “delighted” - they want things to be easy.
We’ve written a separate article on what NPS doesn’t tell you. In short: scores show you the direction; open questions show you what to do.
Seven rules for a good survey
- Short. Three to five questions, under two minutes. Every extra question lowers the completion rate.
- Timely. The same day or the day after the experience. After a week, customers remember a general impression, not the moments.
- One question at a time. “How friendly and quick were the staff?” is two questions. If the answer is 3, you don’t know which one it refers to.
- No leading questions. “How much did you enjoy our new menu?” assumes they did. “What do you think of the new menu?” doesn’t.
- The same scale, always. If you change the scale or wording from one round to the next, you lose comparability over time.
- At least one open question. “What would have made your experience better?” This is where the most valuable information is.
- The customer’s channel. SMS or WhatsApp for customers who communicate that way, email for others, a QR code on the way out. One link, no login.
A five-question model
1. How likely are you to recommend us to a friend or relative? (0–10)
2. What is the main reason for your score? (free text)
3. How easy was it to [book / get to us / get done what you needed]? (1–7)
4. What would have made your experience better? (free text)
5. May we contact you for more details? (yes / no)
How to read the answers
- Always report the number of responses. An NPS of 80 from 10 responses and one of 80 from 200 responses don’t mean the same thing. On small samples, scores jump from month to month without anything having changed.
- Read the comments, not just the average. Tag each comment with the moment of the journey it refers to. After two or three months, the patterns become clear.
- Check anomalies. A score of 1 next to an enthusiastic comment is most likely an input error, not an unhappy customer. Note it, don’t ignore it, and don’t build conclusions on it.
- Look at the distribution. A good average can hide a group of customers - a particular service, day or shift - with a much weaker experience.
Close the loop
The most neglected part: what happens afterwards. Customers who gave a low score and agreed to be contacted get a phone call within a few days. And once a quarter, tell all your customers what you’ve changed thanks to their feedback. It’s the best argument for them to answer next time.
Personal data
A survey collects personal data, sometimes sensitive (especially in a clinic). State clearly what you use the answers for, ask for consent before contacting customers again and don’t keep data longer than you need. Your privacy policy should cover feedback too.
A survey isn’t a form. It’s a promise that you’re listening - and it’s worth keeping.
