CSAT: how to measure customer satisfaction and what to do with the result

What CSAT is

CSAT, or customer satisfaction score, measures how happy a customer is with one specific experience with your company. It is a moment: a call with a salesperson, an onboarding, a support interaction, an order fulfillment. This is what sets CSAT apart from broader loyalty measures such as NPS, which asks about willingness to recommend the company in general. In practice, the difference is clear: if a customer gives a five in CSAT after a successful support call, but a six out of ten in NPS because they have been waiting six months for a promised integration, both results are true and they measure different things. CSAT tells you that a given interaction worked, while NPS shows that the relationship as a whole needs attention. Blending these two measures into a single metric leads to conclusions that explain nothing.

The typical question is: "How satisfied are you with today's interaction with our team?" and the answer is given on a 1-5 scale, less often 1-7.

How to calculate CSAT

The CSAT score is calculated as the percentage of positive responses, most commonly 4 and 5 on a five-point scale. The formula looks like this:

CSAT = (number of 4 and 5 ratings / total number of ratings) × 100%

Example: if you collected 80 responses in a given month, of which 56 were fours and fives, CSAT is 70%. Simple. The difficulty starts with interpretation, because the same score can mean something completely different depending on whom you asked and what you asked about.

Two methodological decisions change the result more than anything else. First: when you send the survey. If you ask about satisfaction a week after the conversation, you are measuring memory, not experience. Second: whom you ask. Responses from people who chose to fill out the survey on their own are systematically biased, usually toward extreme ratings. That is why in practice it is worth asking right after the event and aiming for a high volume of responses, even at the cost of depth in a single survey.

It is also worth distinguishing transactional CSAT from relational CSAT. The first concerns a single event, the second an overall assessment of the relationship. Mixing them in one report is the most common mistake I see among clients. The score drops, but nobody knows whether it is the effect of one support mishap or the effect of the product no longer keeping up with the customer's needs.

What really drives the CSAT score in a B2B company

In B2B, the rating of a single experience is usually influenced by a few things that can be anticipated and planned for. The first is response time. Imagine Anna, an operations director at a manufacturing company, writing on Monday morning about an error on an invoice worth 40 thousand. She gets a reply on Thursday, from someone in a different department who does not know her orders and asks her to send the customer number. In the survey she gives a 2, even though the error was eventually corrected. The second is predictability. If the customer does not know when they will get a response or who will handle their case, satisfaction drops, even if the final outcome is good.

The third factor is the competence of the front line. In B2B companies where sales and support are separate, the client often reaches someone who does not know the history of their orders. This immediately lowers the score, because the client has to repeat what they already said. The fourth is consistency between promise and delivery. A salesperson promises implementation in two weeks, implementation takes four, CSAT drops, even though neither team did anything objectively wrong.

There is another factor that is discussed less often: how the client perceives their position in the relationship. Clients who feel treated as major accounts rate higher than those who feel like one of many. In B2B this is often a matter of small details: whether the account manager knows the client's company name, whether they remember the previous conversation, whether they respond within the time they themselves declared.

On the other hand, there are factors that affect CSAT less than is commonly believed. Price is rarely the main reason for a low rating of a single interaction, though it can be a reason for a client leaving. The same goes for overall product quality: if a client calls with a specific problem, they rate the conversation, not the product. This matters, because many companies try to improve CSAT with price cuts, which usually does not work.

What CSAT results are good and what should concern you

There is no single number that would be a universal threshold for all industries. The ranges differ between sectors, sales models, and types of contact. Experience shows that in service companies, where the client contacts rarely but about important matters, expectations for service quality are higher than in product companies, where contact concerns simple technical issues. This is a market observation, not a research result, so treat it as a reference point, not a hard threshold.

Instead of looking for a single number, it is worth looking at three things. The first is the trend over time. If CSAT falls for three quarters in a row, that is a signal, regardless of whether the current level is 60% or 85%. The second is the spread between teams or channels. If one account manager has a CSAT of 90% and another 55%, the problem lies in a specific person or in how their cases are assigned. The third is the share of the lowest ratings. Even with high CSAT, if 10% of responses are ones, it is worth examining those cases separately.

It is also worth remembering that high CSAT is not a goal in itself. A company can have 95% satisfied clients and at the same time lose its largest accounts, because those who leave simply stopped answering surveys. That is why CSAT is best read together with other data: client retention rate, number of complaints, and response time. If you want to check how this data looks in your company and what follows from it, it is worth starting with a marketing audit, which organizes measurement and shows where you are losing clients.

How to improve CSAT in B2B

To improve CSAT, you need to change what the client actually experiences in the interaction. Another survey will not do that. Below is a set of actions that in practice deliver the greatest effect in B2B companies.

1. Determine exactly what you are measuring

Choose one or two events that have the greatest impact on the client relationship. Most often these are the first sales contact, implementation, and handling a request. For each of them, write a separate question and calculate CSAT separately. A single company-wide metric is convenient for reporting but useless for action.

2. Shorten the time between the event and the question

A survey sent within a few hours of contact gives answers based on actual experience. One sent after a week measures only a general impression. If you cannot ask right away, it is better to ask a smaller sample, but quickly.

3. Respond to low ratings within 24 hours

A customer who gave a low score usually expects someone to reach out. If you don't, that score hardens into an opinion about the whole company. A short message is enough: thank them for the feedback and tell them specifically what you plan to do. This often lifts satisfaction more than solving the problem itself.

4. Collect comments, not just numbers

A numeric score tells you something is off. A comment tells you what exactly. Even if only 20% of customers leave a description, that is usually enough to spot a recurring pattern. Read them regularly, not once a quarter.

5. Train your team on your own data

Instead of generic customer service training, show your team specific cases from your own surveys. This works better because it deals with real situations rather than abstract rules. If you want to connect CSAT improvement with a broader change in how you reach B2B customers, see what a B2B marketing strategy based on data rather than assumptions looks like.

6. Measure the effect of changes, not just the metric itself

After making changes, compare CSAT before and after, but under the same conditions: the same type of contact, the same channel, a similar customer group. Otherwise it is easy to credit the improvement to something that did not cause it. If you want to know how to measure customer satisfaction in a way that actually lets you draw conclusions, sticking to one methodology over a longer period is key.

CSAT is useful when you treat it as a tool for finding problems, not as a number to report to management. The metric alone will not improve service. What improves it is the decision about what you change in the process, and consistently checking whether the customer noticed.