A parcel turns up and the slip asks how satisfied you are with the delivery. A broadband fault gets fixed and a text afterwards asks how easy it was. Once a year the bank wants to know whether you would recommend it to a friend. Same subject, three different numbers, and all of them get filed under customer satisfaction.
Measuring customer satisfaction almost always comes down to asking, and asking in exactly the same way every time. Three metrics have become the standard for that. The NPS (Net Promoter Score) asks about recommending you, the CSAT (Customer Satisfaction Score) about satisfaction with one experience and the CES (Customer Effort Score) about the effort a request took.
You measure customer satisfaction by asking the same short question after every important contact, turning the answers into a metric such as NPS, CSAT or CES and following that figure month by month. At empirio.ai, an online survey tool made in Germany, matrix, ranking and NPS question types are free. By the end of this guide you will know which of the three suits your situation.
📌 Key points at a glance
- Measuring means the same question on the same scale, every round.
- NPS measures loyalty, CSAT satisfaction, CES effort.
- NPS looks at the whole relationship, CSAT and CES at single contacts.
- Your own previous score says more than another company’s benchmark.
- You learn the most by combining two of the three.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — with hosting in the EU.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
Measuring customer satisfaction: the four routes
There are four ways to find out how satisfied your customers are, and only one of them produces a figure you can line up month after month. The other three still earn their place, because they explain what sits behind that figure.
| Route | What it tells you | What it misses |
|---|---|---|
| Survey with a metric | a score you can compare over time | the reasoning behind it |
| Open question in the survey | reasons in your customers’ own words | comparability from round to round |
| Complaints and reviews | the loudest problems | the quietly unhappy |
| Behaviour such as repeat orders | what customers actually do | the why |
The metric is the frame and the open question fills it in. Complaints and reviews arrive uninvited, so they never cover the whole customer base, only the two edges of it. Behaviour is the most honest of the four routes and the most silent one, because an order that never comes back gives no reason.
All three metrics in this guide belong to the first route. Where they differ is in what they ask, when they ask it and how the answer is worked out.
What NPS, CSAT and CES have in common
The three questions sound quite different, yet they are built on the same idea: a short question, worded the same way every time, that lots of people can answer in seconds. What comes out is a number you track over time. That gives them five things in common:
- One core question: each metric comes from exactly one question on a fixed scale.
- An open follow-up: only the reason underneath turns the number into something you can act on.
- Consistent measurement: wording, scale and timing have to stay put, or the comparison falls apart.
- Trend before benchmark: your own previous score tells you more than other companies’ figures.
- Limits to interpretation: none of the three explains on its own why it goes up or down.
For your own staff there is a separate version of the NPS, the Employee Net Promoter Score. It asks whether employees would recommend the organisation as a place to work.
Note
Alongside NPS, CSAT and CES you will often meet the Customer Satisfaction Index, or CSI. There is no fixed formula behind it. It is a weighted average: several satisfaction questions are scored separately and then combined according to how much each one matters to customers. Anyone who has not measured those weights ends up guessing them, and that is where the index usually comes unstuck in practice.
NPS, CSAT and CES compared
The differences lie in the detail: the question, the scale, the sum and the right moment to ask. Once you have seen the three side by side, you stop mixing them up.
Two tables keep the levels apart. The first shows what gets asked and when, the second what comes out at the end.
Question and timing
The question decides what a metric is able to capture at all, the timing decides how sharp the answer turns out. The two belong together and get settled once.
| Metric | Typical question | When you ask |
|---|---|---|
| NPS | How likely is a recommendation to a friend? | on a fixed rhythm, such as quarterly |
| CSAT | Overall, how satisfied were you with us? | straight after a purchase, delivery or appointment |
| CES | How easy was it to get your issue resolved? | straight after an order, return or support case |
The NPS looks forward and takes in the whole relationship, while CSAT and CES look back at one particular event. The rhythm follows from that: a recommendation question after every contact would be pointless, and a satisfaction question once a year just as much.
Calculation and range
The sum takes two lines for each of the three, yet the results sit on scales that cannot be held against each other. Every figure stays in its own range.
| Metric | Calculation | Range |
|---|---|---|
| NPS | % promoters (9 and 10) minus % detractors (0 to 6) | minus 100 to plus 100 |
| CSAT | answers of 4 and 5 divided by all answers, times 100 | 0 to 100 per cent |
| CES | sum of all ratings divided by the number of responses | 1.0 to 7.0 |
The scales are part of the sum and are not a matter of taste: 0 to 10 for the NPS, 1 to 5 for the CSAT, 1 to 7 for the CES. You can work out all three with our free calculators: calculate the NPS, calculate the CSAT and calculate the CES.
Strengths, weaknesses and origin
Each of the three is strong for its own purpose and weak for everything else. Taking them in turn:
- The NPS gives one figure for loyalty overall, but does not show where the problem lies. Fred Reichheld introduced it in the Harvard Business Review in December 2003 under the title The One Number You Need to Grow.
- The CSAT pinpoints the touchpoint that is letting you down, yet says little about loyalty. It has no single author, because the satisfaction question is older than its abbreviation.
- The CES finds friction in processes and sees only the process, not the product. It goes back to Stop Trying to Delight Your Customers by Dixon, Freeman and Toman in the Harvard Business Review of July 2010.
Which metric to use when
It is rarely a case of one or the other. Each metric answers a different question and does its best work at a different point:
- Go for the NPS when you want to keep an eye on customer loyalty as a whole and follow it over months.
- Go for the CSAT when you want one particular touchpoint rated, such as a delivery, an appointment or an event.
- Go for the CES when you want to improve a process and need to find out where customers get held up.
In practice they complement each other well: CES and CSAT straight after important contacts, the NPS at longer intervals for the overall picture. When the NPS drops, the other two will often already show you where the trouble started. How to put a survey like that together is covered in our guide to customer surveys.
Business customers work differently from consumers
In B2B you have fewer customers but several people inside each one, and only one of them signs the renewal. What counts there is not only the average but also who answered: the buyer in procurement and the person using the service every day often rate it quite differently. In B2C it works the other way round, because a single answer says almost nothing and only the trend does. An NPS built on eight answers is worth a phone call in B2B; in B2C it is just a number.
Create a survey for free
With empirio.ai you can create a modern online survey in minutes — with hosting in the EU.
- AI-built survey
- Adjust by drag & drop
- Real-time analysis
What a tool has to do for the measurement to hold
A tool for satisfaction measurement needs four things: it has to build the scale exactly as the metric defines it, repeat the same survey unchanged as often as you like, count the answers group by group and make the free text readable. Everything beyond that is comfort.
The first one sounds trivial and trips up the most people. If your tool can only build the NPS scale from 1 to 10, you do not have an NPS, you have a ten-point scale whose results match nothing that has ever been published. So before the first round, check that it knows 0 to 10, 1 to 5 and 1 to 7 as separate question types.
Repeating and comparing
A satisfaction measurement earns its keep on the second round. In practice that means setting the survey up once, reusing it for each wave and putting the results side by side without touching a single question. At empirio.ai the comparison mode starts with Lite, cross tabulations and filters start with Plus, and evaluating free text with AI also starts with Plus (as of October 2026). Those limits are worth saying plainly: the free plan gives you the results and the charts, but not the comparison between two waves and not the AI reading of open answers. For a first round the free plan with 50 participants per month is enough, and you can run an online survey at no cost before you settle on a metric for good.
Once the second wave is due and you want both rounds lined up in one view, set up an account and keep them in the same project.
Data protection: when you need a contract
A customer survey processes personal data as soon as a company, a charity or a public body stands behind it. The provider then processes that data on your behalf, and Article 28 of the UK GDPR covers exactly that situation: processing by a processor has to be governed by a contract that binds the processor to the controller and sets out subject matter, duration, nature and purpose of the processing. At empirio.ai the data processing agreement is included in the free plan, so there is nothing to buy before the first survey goes out.
Common mistakes when measuring customer satisfaction
The four mistakes below have one thing in common: they only show up at the second measurement, and by then the first one is already lost.
The question gets reworded along the way
A metric is only worth something next to itself. Change the wording, the scale or the timing between two waves and you have two measurements instead of a trend. So write the question once, in peace, have two colleagues read it back to you and then freeze it.
Asked so often that nobody answers any more
Survey fatigue comes less from long questionnaires than from repetition with no visible consequence. Decide in advance how often one person may be asked in a year, and put a line in the invitation about what came of the last round. Reporting what changed gives people a reason to answer again.
Other companies’ benchmarks instead of your own history
Industry figures from published reports come from other samples, other scales and other moments in time. An NPS of 30 can be excellent in one market and disappointing in the next. Compare against your own last score first, and against everything else only after that.
Measured, then nothing changed
A metric that produces no task only costs attention. After each wave, take the reason that comes up most often in the free text, derive exactly one change from it and check at the next round whether it worked.
Careful
The finer you slice the results, the quicker single answers become traceable to a person. Break an evaluation down by site, contract type and age group and you sometimes have two people left, both of whom recognise themselves. So only evaluate subgroups above a minimum number of responses that you set in advance, and name that threshold in the invitation itself.
Conclusion
Choosing between NPS, CSAT and CES is the smaller decision. The bigger one is whether you keep asking the question unchanged for years and whether every round produces a change. Start with the metric that matches your most frequent customer contact, and add the second one only once the first has a clean run of history.
A metric without a previous score is not a measurement, it is a snapshot.
What to read next
- Answers already in and ready to be analysed? Evaluating a questionnaire
- Response rate lower than you hoped? Tips for online surveys
- Measuring satisfaction in your own team? Employee satisfaction surveys
Want to survey your customers with NPS, CSAT or CES?
With empirio.ai you can build your customer survey in a few minutes and watch the analysis fill up as the first answers arrive.