Home › Blog › NPS and CSAT
Measurement

NPS and CSAT: what each one measures and why you need both

NPS measures the relationship with the brand. CSAT measures the moment you just delivered. Most measurement programs in retail fail because they treat the two as the same thing.

The problem

Two different questions, two different decisions

It is common to see a dashboard with an NPS of 62 and a CSAT of 91% and conclude that everything is fine. The problem is not the numbers: it is that neither one is connected to a concrete decision for next week.

NPS and CSAT are not two ways of measuring the same thing on different scales. They measure different things, they are answered at different moments and they serve different decisions. When they are used interchangeably, you usually end up with two indicators that move up and down without anyone being able to explain why.

Here is the practical difference, how each one is calculated, how often to ask, and what to do in the most interesting case of all, which is when the two do not tell the same story.

CSAT

Satisfaction with one specific moment

CSAT, or Customer Satisfaction Score, measures how satisfied the customer was with something that just happened: a purchase, a delivery, a size exchange, a conversation with customer service.

The question is asked right after the moment you want to evaluate, and it names that moment explicitly: “How satisfied were you with your purchase today?” or “How satisfied were you with the service you received?”. Naming it matters, because the customer answers about whatever they understand you are asking.

The most common scale runs from 1 to 5, from very dissatisfied to very satisfied. There are also 1 to 7 and 1 to 10 versions. Any of them works as long as you do not change it halfway through the year.

The standard calculation is top two box: add up the 4 and 5 answers, divide by the total number of answers and express the result as a percentage.

CSAT = (4 and 5 answers ÷ total answers) × 100

Some teams prefer to report the average of the scale. Both are valid, but they are not comparable with each other. If you change method, your historical series stops being useful even though the indicator keeps the same name.

CSAT is good at one thing and bad at another. It is good at locating where the operation breaks, because it can be cut by store, by associate, by channel and by type of interaction. It is bad at anticipating whether the customer will come back: someone can be satisfied with today's delivery and never buy from you again.

NPS

Willingness to recommend the brand

NPS, or Net Promoter Score, measures something different: how willing the customer is to recommend the brand. It does not ask about a moment, it asks about the whole relationship.

There is a single question: “On a scale of 0 to 10, how likely are you to recommend our brand to a friend or family member?”. It is always worth pairing it with a second, open question: “What is the main reason for your answer?”. The score tells you how much, the open answer tells you what to do.

Answers are grouped into three: promoters are those who answer 9 or 10, passives those who answer 7 or 8, and detractors everyone who answers 0 to 6.

The calculation ignores passives. You subtract the percentage of detractors from the percentage of promoters, and the result runs from -100 to 100. It is reported as a number, not as a percentage.

NPS = % promoters − % detractors

The detractor band tends to look surprisingly wide. That is deliberate: in practice a 6 is not a moderately happy customer, it is someone who already has a reason to go elsewhere next time.

NPS is good for tracking the health of the relationship over time and for comparing stores or regions against the same yardstick. It is bad for diagnosis: a falling NPS does not tell you whether the problem is the assortment, the checkout line or the returns policy.

The difference

Moment versus relationship

The shortest way to separate them is this: CSAT evaluates what just happened, NPS evaluates what the customer thinks of you in general.

Everything else follows from that, including how often to ask. CSAT can be asked after every relevant interaction without becoming a nuisance. NPS cannot: if you send it after every ticket you burn it, and you also stop measuring the relationship and start measuring the last day. In retail with frequent purchases, two or three times a year per customer is enough.

CSATNPS
What it measuresSatisfaction with one momentWillingness to recommend the brand
Scale1 to 5, also 1 to 7 or 1 to 100 to 10
ResultPercentage, from 0% to 100%Number, from -100 to 100
When to askRight after the interactionTwo or three times a year
What it is forFinding where the operation failsTracking the health of the relationship
Where it falls shortIt does not anticipate whether the customer returnsIt does not say what needs fixing
How they complement each other

The interesting part is when they disagree

If both indicators rise together, the reading is easy and needs no article to explain it. The value of having both shows up precisely when they contradict each other, because each combination calls for a different action.

High CSAT and high NPS

The operation delivers and the brand convinces. The job here is to sustain the level and give promoters an easy way to say so in public.

High CSAT and low NPS

Every interaction goes well, but the customer does not recommend you. The problem is not in the operation, it is in the proposition: price, assortment, convenience or the reason to prefer you. This is the scenario most often mistaken for success.

Low CSAT and high NPS

The customer likes the brand and the operation is failing them. It is the most recoverable of the four scenarios and the most expensive to ignore, because a promoter runs out of patience too.

Low CSAT and low NPS

This is not an execution problem. Tweaking the checkout script will not help here: the whole proposition has to be reviewed before measuring again.

None of these four readings is possible with a single indicator. That is why the decision is not choosing between NPS and CSAT, but being clear about which question each one answers.

In practice

Four mistakes that keep repeating in retail

Sending NPS after every purchase

It turns a relationship question into a transaction question, saturates the customer and drags down the response rate of every survey you send afterwards.

Changing the calculation halfway through the year

Moving from top two box to the average, or from a 5-point scale to a 10-point one, breaks the series. The new indicator is not comparable with the old one even though it has the same name.

Reading only the people who answer

The angriest customer often does not reply, they simply leave. If the response rate is low and the number is going up, you may only be measuring your fans better and better.

Measuring without closing the loop

A detractor who rated you poorly and got no response ends up worse than before the survey. The value of measuring is not in the number, it is in what happens next with whoever rated you poorly.

Where it changes

What happens when the survey lives next to the purchase

Almost every survey tool has the same limit: it knows what the customer answered, but not what the customer did afterwards. The score sits in one tool and the sale in another, and connecting them becomes a project every time someone asks.

At Guper the survey comes out of the same platform that already holds the purchase. That changes three concrete things.

The survey is triggered by the purchase

It goes out over WhatsApp after the ticket, with the store, the date and the customer already identified. No files to cross-reference to know which purchase each answer belongs to.

The answer stays on the customer profile

Whoever answers on WhatsApp or works the sales floor sees the rating next to the purchase history and the balance, not in a separate tool.

The breakdown reaches the store and the question

Month-by-month evolution, CSAT question by question, performance by store, and the themes detractors repeat most.

And the question almost no survey tool can answer becomes available: did March's detractor buy again? Because the sale lives in the same platform, that follow-up stops being a project and becomes a query.

Want to see how this is measured in your operation?

In the demo we review what you measure today and what can be connected to sales.

Book a demo