A familiar picture from everyday shop life: your review profile shows almost five stars, yet your support inbox is full of irritated questions about delivery. Review profiles show the mood of the people who choose to write. Which tickets frustrate customers and which process costs them time only becomes visible once you measure it yourself.
How can you measure customer satisfaction?
Measuring customer satisfaction means asking customers right after a specific interaction how satisfied they were and turning their answers into a metric. In e-commerce you use three values for this. CSAT measures individual experiences such as a resolved ticket, NPS measures loyalty to your brand and CES measures effort, for example during a return. The results are most meaningful when you ask in the same channel where the interaction took place.
How much retailers differ at individual touchpoints is shown by the 2026 retail study of the American Customer Satisfaction Index. It is based on 31,293 surveys across all retail sectors, and online retailers score an average of 79 out of 100. For the first time, the ACSI measured delivery quality (83 points) and ease of making returns (80 points) separately. Between individual retailers, these two values differ by 15 to 20 points. An overall score hides exactly the differences you can work on. The data comes from the US. For your shop this means delivery, returns and service each need their own measuring point.
Which metrics measure customer satisfaction?
There are three established metrics for customer satisfaction: CSAT, NPS and CES. Each one answers a different question. That is why you choose the right customer satisfaction KPI based on the moment you want to evaluate.
| Metric | Typical question | Scale | Calculation | What it measures |
|---|---|---|---|---|
| CSAT (Customer Satisfaction Score) | “How satisfied were you with our help?” | 1 to 5 | Share of answers with 4 or 5 | Satisfaction with a single interaction |
| NPS (Net Promoter Score) | “How likely are you to recommend us?” | 0 to 10 | Percentage of promoters (9 and 10) minus percentage of detractors (0 to 6) | Loyalty to the brand |
| CES (Customer Effort Score) | “How easy was it to resolve your request?” | usually 1 to 7 | Share of answers with 5 to 7 | Effort in a process |
The rule of thumb: CSAT for service, CES for processes, NPS for the brand. Ask only one of these questions per survey. That way it stays clear what the answer refers to.
On top of that come indirect signals such as repeat purchase rate, return rate and the share of complaints. These values show you the effect, but only a direct question reveals the cause. Operational service values such as response time and resolution time belong in a separate dashboard. How to calculate them is covered in our overview of the most important KPIs in customer service.
How can you measure CSAT?
You measure CSAT with a single question right after a completed interaction. A scale from 1 to 5 is common. In WhatsApp or live chat the levels appear as buttons, in email as clickable stars.
The formula: answers with 4 or 5 divided by all answers, times 100. A worked example: 200 customers answer and 164 of them choose 4 or 5. Your CSAT is 82%.
To make the value meaningful, watch three details:
- Calculate with the share of satisfied answers. An average of 4.1 stars sounds good and hides whether one in five people gave a 1.
- Measure CSAT per ticket category. An overall score mixes a quick invoice request with a complaint about a broken product.
- Separate service and product. “How satisfied were you with our help?” rates your team. “How satisfied are you with your order?” rates the goods and the delivery.
Add an open question as soon as someone chooses 1 to 3. A line such as “What could we have done better?” gives you the reason. The reason is more valuable to your team than the number itself, because it shows directly which process is causing trouble.
How can you measure NPS?
You measure NPS with the recommendation question on a scale from 0 to 10. Fred Reichheld introduced the method in the Harvard Business Review in 2003. It divides all answers into three groups:
- Promoters give 9 or 10 and would recommend you.
- Passives give 7 or 8 and are satisfied but not enthusiastic.
- Detractors give 0 to 6 and are dissatisfied.
The formula: percentage of promoters minus percentage of detractors. A worked example: of 300 answers, 150 come from promoters (50%) and 60 from detractors (20%). Your NPS is 30. The value ranges from minus 100 to plus 100 and is stated without a percentage sign.
Passives only count towards the total number of answers. They are still interesting for you, because small improvements can turn them into promoters. What you actually do with the answers from detractors and passives belongs to the question of how to raise satisfaction. For that, we have collected 10 measures for higher customer satisfaction.
Ask the NPS question after the overall experience, a few weeks after delivery. By then the person has tried the product and can rate the brand as a whole. Right after a support ticket, NPS mainly measures their mood about the problem that has just been solved.
When is the Customer Effort Score the better choice?
CES asks how much effort a request took, which makes it a good fit for processes such as returns, exchanges or address changes.
Matthew Dixon, Karen Freeman and Nicholas Toman made the metric known in the Harvard Business Review in 2010. It was based on a study of more than 75,000 customers. Their finding: loyalty comes mainly from quick, simple solutions. Extra attention beyond what customers expect makes little difference. The authors consider CES a better indicator of loyalty than CSAT and NPS.
In a shop it looks like this: a customer registers a return. Right afterwards you ask: “How easy was it to register your return?” The scale varies by tool, in this example it runs from 1 (very difficult) to 7 (very easy). If many answers land on 1 to 3, look at the process first. Possible causes are a returns portal that cannot find the order number or a label that only arrives after a second message.
For product advice and complaints, CSAT measures more directly. What counts there is the feeling of having been treated well. A customer whose parcel got lost rates your team's friendliness and solution. Effort is secondary in that moment.
When and in which channel should you ask?
The best measurement happens right after the moment in question and in the channel the person is currently using. A survey by email after a WhatsApp chat forces customers to switch channels.
| Touchpoint | Metric | Timing | Channel |
|---|---|---|---|
| Support ticket resolved | CSAT | right after resolution | Channel of the ticket (email, WhatsApp, chat) |
| Return or exchange registered | CES | right after registration | Channel of the registration |
| Order delivered | CSAT on product and delivery | one to three days after delivery | WhatsApp or email with consent |
| Overall experience | NPS | two to four weeks after delivery, at most once per quarter | WhatsApp or email with consent |
| Complaint closed | CSAT plus open question | after resolution, reviewed personally | Channel of the complaint |
Track the response rate alongside the score from day one. If you only count the answers you receive, you overlook everyone who does not respond at all. If the rate drops, check the length of the question, the timing and the channel first. For surveys after delivery, WhatsApp is a good choice because the question lands in the chat customers already use. The prerequisite is a WhatsApp opt-in.
If you want to bring CSAT in tickets and surveys after delivery together in one system, we are happy to show you how it works with your own setup. Book a demo and see how armincx by Chatarmin measures CSAT directly in the ticket.
How do you set up a customer satisfaction survey in 5 steps?
A good survey is short, arrives at the right time and stores its result in the customer profile. Here is how to go about it:
- Ask one question per survey. The open question only follows when someone gives a low score.
- Make the answer a single click. Buttons in WhatsApp, stars in email, smileys in the chat widget.
- Save the score in the customer profile. At the next contact your team can see immediately how the last experience went.
- Hand low scores over to a person. Anyone who gives 1 to 3 gets a personal reply within one working day.
- Analyse weekly and decide monthly. Individual outliers tempt you into hasty process changes.
A survey after delivery can look like this: “Hi Lena, your order arrived two days ago. How satisfied are you with it?” The answer options appear below as buttons. Keep the message short and address the person by name.
With Chatarmin, which more than 500 brands use for WhatsApp marketing, you send the survey after delivery as a message with buttons. Every click automatically adds a tag to the customer profile. Through the Klaviyo integration, dissatisfied customers end up in their own segment, which you can address in a targeted and personal way.
In customer service, armincx asks for CSAT after the ticket is resolved. Low scores go straight to your team so that someone takes care of them personally. How to create a survey in WhatsApp step by step is explained in our guide to WhatsApp surveys.
How do you analyse the results?
One number per month is enough for management, but for improvements you need breakdowns. These four comparisons show you where to start:
| Breakdown | What you compare | What you learn from it |
|---|---|---|
| Ticket category | Where is my order, return, cancellation, invoice, complaint | Which process produces the most poor scores |
| Channel | Email, WhatsApp, Instagram, chat | Where answers take too long or information is missing |
| Resolution type | AI, workflow, person | Whether automated answers are received as well as personal ones |
| Time period | Normal weeks versus Q4 and Black Friday | Whether quality drops under pressure |
An everyday example: the overall CSAT stays stable, only the category “Where is my order” slips in November. Then it pays to look at your shipping information first, before you work on your team's tone. Proactive shipping updates address exactly this point.
The channel breakdown shows you whether one channel performs worse than the others. If CSAT only drops in one channel, you can see exactly where your team needs more capacity or better templates. Always compare channels with a similar number of answers.
Closed quickly and satisfied are two different things. Workflow-automated tickets close after a median of one day, manually handled ones after three days. A wrong answer in ten seconds still creates a follow-up ticket, which is why CSAT belongs right next to resolution time.
The distribution of tickets also shows you where this matters most: 63% arrive by email, 15.4% via Instagram and 11.2% via WhatsApp. All figures come from the Chatarmin benchmark on AI in customer service covering 131 shops and 2.9 million resolved tickets.
What should you avoid when measuring customer satisfaction?
The riskiest mistakes are review invitations sent only to satisfied customers and surveys without consent. On top of that come mistakes that cost you meaningful results:
- Sending review invitations only to satisfied customers. Many flows ask for CSAT first and only send the review link for 4 or 5 stars. Trustpilot's guidelines for businesses prohibit exactly this kind of selective invitation and require all customers to be invited in the same way. Your internal survey may forward dissatisfied customers to support, but everyone still gets the review invitation.
- Rewarding reviews with vouchers. According to the same guidelines, Trustpilot does not allow incentives such as discounts, prize draws or refunds for a review.
- Sending surveys by email without consent. In 2018, Germany's Federal Court of Justice ruled (case VI ZR 225/17) that a satisfaction request in an invoice email is not permitted without prior consent (report by LTO, in German). Obtain consent at checkout or with the WhatsApp opt-in and clarify the details with your data protection adviser.
- Asking after every contact. Set a pause per person, for example at most one survey after delivery every 30 days.
- Using CSAT as a bonus metric for individual agents. Surveys then tend to end up with the easy cases. Use the value for processes and coaching.
- Overrating small samples. With 15 answers, a single rating shifts CSAT by almost 7 percentage points.
If AI answers part of your tickets, measure its CSAT separately from your team's answers. That way you see which requests the AI handles well and where customers prefer to talk to a person. Which requests suit AI and where a person takes over is covered in our guide to AI in customer service.
Frequently asked questions about measuring customer satisfaction
Which metrics are there to measure customer satisfaction?
The three standard metrics are CSAT for individual interactions, NPS for willingness to recommend and CES for effort. Indirect values such as repeat purchase rate, return rate and complaint share add to the picture.
What does CSAT tell you?
CSAT shows what percentage of respondents were satisfied with a specific interaction, usually as the share of answers 4 and 5 on a five-point scale. It rates a single moment such as a resolved ticket, which makes it a particularly good fit for customer service.
What is a good CSAT score?
That depends on the scale, the industry and the ticket mix. Compare yourself with your own baseline first and track the development per ticket category.
How often should you measure customer satisfaction?
You measure CSAT and CES continuously after every relevant interaction, and NPS at most once per quarter per person. Analysing weekly pays off, and you make decisions on a monthly basis.
Do you need consent for a customer satisfaction survey by email?
It depends. For a satisfaction request in an invoice email, Germany's Federal Court of Justice required prior consent in 2018 (case VI ZR 225/17), so with consent you are on the safe side.
Can I measure customer satisfaction via WhatsApp?
Yes. After a support chat you ask directly in the ongoing conversation, and for surveys after delivery you need a WhatsApp opt-in.
Is NPS better than CSAT?
No. Both measure something different: CSAT a single interaction, NPS loyalty to your brand.
Conclusion: measure satisfaction where it happens
Measuring customer satisfaction in an online shop works best with three metrics at the right time: CSAT after the ticket, CES after processes such as returns, NPS after the overall experience. Ask in the channel of the interaction, save the score in the customer profile and analyse by ticket category.
Dissatisfied customers reach a person, and everyone gets the review invitation. Which tools you need for surveys and analysis is covered in our overview of customer feedback software. Start with a single CSAT question after every resolved ticket and check after four weeks which ticket category delivers the poorest scores.
