Customer Satisfaction Score (CSAT): Definition & Calculation

Picture a garden centre on the ring road outside Leeds. Since the spring it has offered click and collect: you order compost, bedding plants or a new patio set on the website, then drive round to a bay beside the car park. A QR code on each collection slip leads to one question. In April, 88 per cent of replies are positive. Over the late May bank holiday that falls to 71. Nothing on the website had changed. Two of the yard staff were off sick, and customers stood in the drizzle for twenty minutes to collect a bag of multi-purpose compost.

The Customer Satisfaction Score is a metric for measuring how satisfied customers are with one specific experience, whether that’s a purchase, a delivery or a call to customer services. Shortly after that experience, people rate it from 1 (very dissatisfied) to 5 (very satisfied). Your CSAT is the percentage of them who picked 4 or 5. You put the question to them with a survey tool: at empirio.ai, an online survey tool made in Germany, the matrix, ranking and NPS question types are free. By the final section you’ll have calculated a CSAT yourself, you’ll know which bands to read it against and you’ll recognise the three situations in which it quietly misleads.

📌 Key points at a glance

• CSAT rates one experience on a five point scale, 1 to 5.
• Satisfied means a 4 or a 5, nothing else.
• CSAT = satisfied answers ÷ all answers × 100.
• Results range from 0 to 100 per cent.
• Best asked within hours of a purchase, delivery or support call.

What is the CSAT?

Behind the abbreviation CSAT, short for Customer Satisfaction Score, sits a single question put at a single touchpoint:

“Overall, how satisfied are you with our service?”

Five labelled answers are on offer, running from 1 (very dissatisfied) through dissatisfied, neutral and satisfied up to 5 (very satisfied).

Unlike the NPS, the CSAT has no author and no trademark behind it. Businesses were asking customers whether they were happy long before anyone gave the practice a name. Its reach is well documented, though. When the Corporate Executive Board studied service organisations for an article in the Harvard Business Review in 2010, roughly four in five of them relied on the CSAT as their primary measure of the customer experience.

Its strength is timing. The question isn’t about your brand, your prices or your reputation. It’s about this morning’s delivery, yesterday’s phone call, the collection bay on Saturday. That narrow focus is exactly what a team needs when it is trying to repair one stage of the journey and wants evidence within weeks rather than quarters.

Satisfied and loyal, though, are two different things, and happy today does not mean back tomorrow. The same research found that one in five customers who called themselves satisfied still planned to take their business elsewhere, while 28 per cent of those who were dissatisfied planned to stay put. So read the CSAT as a verdict on an experience and not as a forecast of loyalty.

Satisfied, neutral, dissatisfied: the answer steps

When you analyse the answers, the five steps collapse into three groups:

• Satisfied with 4 and 5: these, and only these, count towards the score.
• Neutral with 3: not counted as satisfied, yet included in the total, so every 3 pulls the result down.
• Dissatisfied with 1 and 2: the answers worth following up personally.

In market research this way of counting is known as top-two-box. The name gives it away: only the two top boxes score. A 3 earns nothing, because a lukewarm experience is not what you are aiming for. People who pick the middle rarely had a disaster, but nothing happened that they would mention over dinner either.

Variations are common. Some firms run seven steps, some use stars, some a row of faces. Any of these is fine as long as the rule about which answers count as satisfied is fixed before the first survey goes out and never touched again. Technically, each one is a rating scale whose points carry labels.

Satisfaction and effort are two separate things, and the Customer Effort Score captures the second one: how much work the customer had to do. Asking both after a support case pays off. Someone whose parcel finally turned up after two chasing emails may give the driver a 4 for satisfaction and still mark the effort down hard.

Working out the CSAT: formula and worked example

The CSAT is the number of 4s and 5s divided by the number of all answers, times 100, which gives a percentage from 0 to 100.

1. Count each step. Note how often 5, 4, 3, 2 and 1 were chosen.
2. Combine the top two. Add the 4s and the 5s.
3. Convert to a percentage. Divide by the grand total, then multiply by 100.

Got your counts? Work out your CSAT with our free calculator, which gives you the mean rating too.

Here is what 100 click and collect customers in Leeds said after picking up their orders:

The 38 fives and 40 fours add up to 78 satisfied customers out of 100, a CSAT of 78 per cent. Their average rating works out at 4.01.

Now imagine a second garden centre with the same 78 per cent but a different crowd: 31 fives, 47 fours, no threes or twos and 22 ones. Its average is only 3.65. Over a fifth of its customers were very dissatisfied, compared with 4 per cent in Leeds. The percentage misses that, the average and a glance at step 1 do not, so our calculator reports both.

Split the CSAT by touchpoint as well. Online ordering, collection, returns and customer services each deserve a score of their own. Averaged together, 94 per cent for the website and 63 per cent at the collection bay produce a comfortable figure that tells nobody where to act.

What is a good CSAT result?

A CSAT of 80 per cent can be excellent in one sector and mediocre in another. Ratings after a purchase tend to be kinder than ratings after a complaint, and hotel guests are harder to please than people waiting for a parcel. Our CSAT calculator therefore offers bands purely as a rough orientation, to help you read the number rather than to pass judgement on it:

• 85 to 100 per cent: very good. Nearly every customer walks away content.
• 75 to 84 per cent: good. Most are happy, though there is still something to gain.
• 60 to 74 per cent: room to improve. Between a quarter and two fifths of customers are not won over.
• 0 to 59 per cent: critical. Something fundamental is broken at this touchpoint.

CSAT results often look flattering. In Britain especially, people are reluctant to mark anyone down, so a polite “satisfied” regularly stands in for a shrug. That is why a slide of five points is worth taking seriously. The line over time at one touchpoint matters most: if satisfaction with collection slips three months running, something has changed, whatever band it still sits in.

Industry benchmarks: why other people’s figures rarely help

Published average scores for the CSAT are easy to find and almost impossible to compare. Four things differ from one source to the next, and each of them shifts the result: the scale (five steps, seven or ten), the counting rule (top mark only, or the top two), the moment the question was asked (at the till or weeks afterwards) and who was asked in the first place. A score drawn only from resolved support cases is bound to sit higher than one drawn from every case.

Someone else’s figure becomes useful only when the source states all four. If one of them is missing, the number is conversation and not a target. Compare three things you control instead: your own score at the same touchpoint last month, your touchpoints against each other, and the score before and after a change you made deliberately.

Putting the CSAT question into your questionnaire

No metric depends on timing quite as much as the CSAT. Ask while the experience is still fresh: on the card terminal once payment has gone through, by text a couple of hours after delivery, or in the email confirming that a ticket is closed. A week later, people are rating a blur of everything that has happened since.

Precise questions get precise answers. “How satisfied were you with the advice from our plant specialist today?” tells you far more than a vague “How satisfied are you with us?”. To cover several stages, repeat the satisfaction question for each one, as our CSAT survey template does. For the order of the questions around it, see our guide on how to structure a survey.

1 to 5 or 1 to 10: the scale the CSAT needs

Five labelled steps have become the standard for the CSAT, and the reason is practical rather than theoretical: five steps can each carry a word, ten cannot. On a ten point scale the person answering has to decide for themselves what a 7 means and where it stops being an 8. That distinction is then thrown away at analysis time anyway, because the answers end up in two groups regardless.

The ten point scale belongs to the Net Promoter Score, which asks a different question and is worked out in a different way. If you track both, keep both scales and resist the urge to blend them. And once you have settled on five steps, or seven, the scale stays as it is, because every change wipes out the comparison with every earlier round.

CSAT question checklist

• It names the experience being rated, not the company in general.
• Every one of the five points has a word label.
• Satisfied always sits on the same side of the scale.
• It goes out the same day whenever possible.
• The rule “4 and 5 count as satisfied” is agreed before launch.
• Low ratings (1 or 2) lead on to a question about the reason.

What a tool for CSAT surveys has to do

A tool for CSAT surveys has to manage four things: show a five step scale with every step labelled, put the question at the touchpoint itself by link, QR code or embedded in a page, follow a 1 or a 2 with an open question, and count the answers separately for each touchpoint. Everything beyond that is comfort.

The first of those sounds obvious and is the one that trips people up most. A tool that cannot label the steps leaves you measuring with bare numbers, and a bare 4 means something different to everyone who picks it. Before the first round, check that the scale runs the same way round in every question too, because a flipped scale produces answers that nobody can correct afterwards.

Then there is data protection. A customer survey processes personal data the moment a company, a charity or a public body sits behind it, and the provider processes that data on your behalf. Article 28 of the UK GDPR states that processing by a processor shall be governed by a contract or other legal act, so the data processing agreement belongs in place before the first invitation goes out.

At empirio.ai, EU hosting and the data processing agreement are part of the free plan, as are sharing by link, the QR code generator and embedding the survey in your own website. Complex logic jumps start with Lite, cross-tabulations and filters for analysing each touchpoint separately start with Plus, and evaluating free text answers with AI also starts with Plus (as of October 2026). For a first measurement at a single touchpoint the free plan and its 50 participants a month are enough. If you want to try the question on the collection slip for a week before committing, you can start an online survey free of charge and decide afterwards whether the CSAT stays on at every stage.

How to tell the setup actually holds

Three trials will do it, and all three happen before the first real invitation. First: answer the question yourself on a phone and time it. Longer than ten seconds and it is too long for a touchpoint. Second: give it a 2 and check that the open question appears. Third: open the analysis after five test answers and make sure you can see the spread across all five steps and not only the percentage.

Once those three trials pass, it is worth hanging the question on every touchpoint. A questionnaire to experiment with takes a few minutes as soon as you set up a free account.

Pros and cons of the CSAT

The CSAT is usually the first metric a business adopts, since it needs no introduction. The flip side of that familiarity is that people read more into it than it holds.

Where the CSAT is strong

• Everyone grasps a percentage of satisfied customers straight away.
• Each touchpoint can have its own score, so weak spots stand out.
• One quick question means more people bother to reply.
• Feedback arrives while there is still time to put things right.
• Shops, websites, training courses and events can all use it.

Where the CSAT is weak

• Being satisfied does not stop a customer from leaving.
• Courtesy pushes results upwards, so they often look better than reality.
• It rates a single moment and ignores the wider relationship.
• Every firm counts differently, which makes outside comparisons shaky.
• Counting only the top two boxes hides whether the rest were 1s or 3s.

Where the CSAT reaches its limits

Think of the CSAT as a photograph taken from one angle. It is accurate about what is in the frame and silent about everything outside it.

Three gaps matter most, particularly once the score starts appearing in targets or bonus schemes.

Loyalty is outside its view

According to the research in the Harvard Business Review, satisfaction and loyalty were only loosely connected. A customer can have a good experience with you and still move to a rival that is cheaper or on the way home. For intentions to return or to recommend, other metrics do a better job.

Silent customers never show up

Only people who reach a touchpoint can answer. If someone abandons their basket at checkout, or visits once and never returns, no satisfaction question ever finds them. A strong CSAT may simply mean that the disappointed customers have already left.

Other firms’ CSATs are not your CSAT

One company counts 4s and 5s on five points, another counts 6s and 7s on seven, a third only counts top marks. Each calls the result CSAT. Figures published elsewhere are therefore at most a vague reference and never a target.

The CSAT judges the moment, not the relationship.

CSAT, CES and NPS: where they differ

Three metrics dominate customer surveys, and each asks something different:

• The CSAT is about satisfaction with a specific experience, so it judges the moment.
• The CES (Customer Effort Score) is about how much work it took to get something done, so it judges the route.
• The NPS (Net Promoter Score) is about willingness to recommend you, so it judges the relationship.

Our comparison of NPS, CSAT and CES sets out when each fits best and how they work together.

Typical CSAT survey mistakes

A CSAT survey can be live in an afternoon and skewed by the end of the week. For the general principles, have a look at our 8 tips for creating the perfect online survey. The three errors below are the ones that crop up most with the CSAT.

Asking late, or asking vaguely

Send the question a fortnight after purchase and the answer covers the fortnight, not the purchase. An unanchored “How satisfied are you with us?” has the same problem. Tie the question to one experience and ask within hours.

Surveying only the tickets that were closed happily

Plenty of helpdesk tools trigger the CSAT only when an agent marks a case as resolved. Customers whose problem never got fixed are never asked, so the score climbs without the service improving at all. Survey every closed contact, the messy ones included.

Moving the goalposts

Five labelled points turn into emojis, top-two becomes top-one, or the neutral option quietly disappears. Each change shifts the result although customers feel exactly as they did before. Decide on the scale and the counting rule once, document both and pass the document on, so that next year’s figures can still be set against this year’s.

A low CSAT: what to do next

You lift a low CSAT in four steps: read the answers on steps 1 and 2 one by one, write out the reason that comes up most often, derive exactly one change from it and fix the date on which you measure again. Anything beyond that usually just postpones the first step.

In that order, reading matters more than calculating. A score of 61 per cent tells you that something is wrong; three sentences from an annoyed customer tell you what. Twenty free text answers are read in a quarter of an hour and save you the meeting about what the cause might be.

The most common reason beats the loudest one

While you read, the sharpest answer stands out most, and that is rarely the one that matters. So count roughly how often the same subject comes up: waiting time, the condition of the goods, getting hold of somebody, being understood. Four matter-of-fact remarks saying the same thing weigh more than a single complaint in capital letters.

One change, not ten

Ten measures on a slide look decisive and never get done. Take the subject that came up most, derive a single change from it that can be finished in four to six weeks, then measure again at the same point with the same question. If the score holds, that was not the cause. That is a result too, and a faster one than any root cause analysis done in advance.

Do not turn the CSAT into a bonus target

As soon as a CSAT score is tied to a bonus, it is the measurement that changes rather than the service. Staff start asking for a good rating, sending the question out after the cases that went well and quietly leaving it out after the difficult ones. The score rises, customers notice nothing, and the number you wanted to steer by is no longer fit for steering. Use the CSAT as a pointer to where you should look, and judge people on what they changed once they had looked.

Conclusion

Few metrics are as quick to set up or as easy to explain as the CSAT, and few are as easy to over-interpret. Treat it as a close-up of one moment rather than a portrait of the relationship. Tie the question to a specific experience, send it promptly, count only the 4s and 5s, ask unhappy customers why and score each touchpoint separately. Then it will show you, week by week, exactly where your customers’ experience needs work!

What to read next

• Planning a bigger survey of your customers? Our guide to customer surveys
• Results in and want more out of them? How to analyse your questionnaire in five steps

Ready to measure satisfaction at every touchpoint?

With empirio.ai you can set up your customer survey in minutes and share it by link or as a QR code on a receipt or collection slip. The CSAT template already includes a satisfaction question for each stage.