The three core customer satisfaction metrics answer different questions. CSAT measures how satisfied a customer was with a specific interaction, Net Promoter Score (NPS) measures how likely customers are to recommend the company, and customer effort score (CES) measures how easy it was to get something done. Support teams should match each metric to the decision it is meant to inform, and not rely on a single number for all three.
In brief: use CSAT to judge individual interactions, CES to find and remove friction in a process, and NPS to track loyalty over time at the relationship level. A workable measurement framework collects each at a different moment in the customer journey, reads them alongside operational data, and treats movement in the trend as more informative than any single score.
What are customer satisfaction metrics, and why do they differ?
Customer satisfaction metrics are survey-based measures of how customers perceive a product or service. They are often discussed as a group, but they are collected at different points and describe different things. CSAT and CES are transactional: they are triggered by an event such as a resolved ticket and describe that event. NPS is relational: it is usually sent on a schedule and describes the overall view of the company.
This distinction explains most of the confusion in practice. A customer can give a high CSAT rating after a pleasant chat and still be a detractor on NPS because of pricing, product gaps or an earlier bad experience. The reverse also occurs. Reading the three together reveals which of those situations an organization is in, whereas reading one alone can mislead.
The broader set of operational indicators is covered in our guide to customer support KPIs. Our earlier article on how to measure customer satisfaction introduces the survey mechanics, and this article builds on it by comparing the metrics directly.
How is CSAT calculated, and what does a CSAT calculator do?
CSAT asks a direct question after an interaction, for example "How satisfied were you with the support you received?", usually on a five-point scale. The conventional calculation states that CSAT equals the number of satisfied responses divided by the total number of responses, multiplied by 100, where satisfied means a rating of 4 or 5. As a worked example, if 180 of 240 respondents choose 4 or 5, the score is 75 percent.
A CSAT calculator performs this arithmetic from the count of responses at each rating and is useful for checking a result by segment, such as channel or contact reason. The same page serves teams searching for a CSAT score calculator, since the output is the same percentage. The definition of satisfied should be fixed in advance, because including 3 as a favorable response raises the score and prevents comparison with earlier periods.
CSAT is most informative for evaluating individual agents and interaction types. Its limitation is that it combines several influences, including the agent, the outcome and the customer's mood, so a movement in the score does not by itself indicate the cause.
How is NPS calculated, and how does a net promoter score calculator work?
NPS asks one question: "How likely are you to recommend us to a friend or colleague?", answered from 0 to 10. Respondents scoring 9 or 10 are promoters, 7 or 8 are passives, and 0 to 6 are detractors. The Net Promoter Score equals the percentage of promoters minus the percentage of detractors, and passives are counted in the total but do not enter the subtraction.
As a worked example, suppose 200 customers respond. If 110 are promoters, 60 are passives and 30 are detractors, promoters are 55 percent and detractors are 15 percent, so NPS is 40. The result is expressed as a whole number between minus 100 and 100, and it is not a percentage.
An NPS calculator applies the same method and also reports the share in each group, which is helpful because two organizations with identical scores can have very different mixes of passives and detractors. People searching for a net promoter score calculator will find the same function there. NPS is the right metric for tracking loyalty over time and for comparing cohorts, but it is a poor tool for diagnosing a particular support process, because the question does not refer to one.
How does customer effort score fit alongside them?
CES asks how easy it was to resolve an issue, typically on a seven-point scale, and is calculated either as the average response or as the percentage of favorable answers. A full treatment of the wording, formula and worked examples is available in our article on customer effort score, and the customer effort score calculator automates both methods.
Its place in the framework is diagnostic. Where CSAT indicates that something went wrong and NPS indicates that loyalty is changing, CES indicates which part of the journey required too much work. It is therefore the metric most closely connected to process redesign, and it is the most natural fit for automated and self-service journeys, where no agent can lift the score through manner alone.
Which metric should a support team use, and when?
The choice follows from the decision at hand. If the decision concerns agent coaching or the quality of a specific response, CSAT is appropriate. If it concerns whether a workflow, a policy or a self-service flow is creating unnecessary work, CES is appropriate. If it concerns whether the customer base is becoming more or less loyal, or how a segment compares with another, NPS is appropriate.
Most organizations benefit from running all three at different frequencies. CSAT can be sent after every closed conversation, CES after defined transactions such as refunds, changes and onboarding steps, and NPS quarterly or twice a year to a sample of the base. Surveying the same customer with all three after one interaction tends to depress response rates and produces overlapping answers.
Smaller teams with limited capacity should begin with CSAT, because it is the simplest to deploy and interpret, and add CES once they have identified a process that merits redesign. NPS can follow when there is enough volume for the trend to be statistically meaningful.
What does a measurement framework for customer experience KPIs look like?
A practical framework has three layers. The first is the outcome layer, where customer experience KPIs such as retention, churn and lifetime value show whether customers stay and what they are worth. These are lagging indicators and are discussed in our article on retention, churn and lifetime value. The second is the perception layer, made up of the survey metrics described above. The third is the operational layer, which includes first response time, resolution rate and cost, and which management can influence week by week.
The value of the framework lies in connecting the layers. When NPS declines, the organization should be able to trace the movement to a segment, then to a contact reason, then to an operational measure such as slow resolution or repeated contacts. This is the practical meaning of a customer service KPI set that is actionable: each indicator has an owner, a review cadence and a hypothesis linking it to the others. For the operational layer in particular, our guide to first response time and the SLA calculator provide supporting detail.
Tagging contacts by reason is the enabling step. Without it, scores cannot be segmented and the analysis stops at the average. Bund AI includes customer insights such as signals and a weekly digest that help surface which topics drive contacts, which can supply the segmentation that survey scores alone lack.
What pitfalls should organizations avoid?
Several recurring errors reduce the value of satisfaction data. The first is survey bias: low response rates and a skew toward extreme opinions mean the reported figure may not represent the typical customer, so the number of responses should always be reported beside the score. The second is gaming, which occurs when agents are rewarded on CSAT alone and begin to coach customers toward high ratings or to avoid difficult cases.
The third is comparing scores across incompatible definitions. A CSAT of 80 percent using a four-and-five threshold is not equivalent to 80 percent using a three-and-above threshold, and external benchmarks often fail to state their method. This article therefore avoids citing industry averages, and organizations should be cautious with any that lack a stated scale, wording and sample.
The fourth is measuring without acting. A survey program that does not result in a documented change loses credibility with customers and staff. Closing the loop on low scores, whether through a follow-up contact or a process fix, matters as much as the collection. A related discussion on improving the underlying experience appears in our article on how to improve customer satisfaction.
When do survey metrics fall short?
Survey metrics are a sample of opinion, and they should not be treated as a complete picture of customer sentiment. Customers who leave quietly without responding are invisible to them, which is why behavioral data such as repeat contacts, cancellations and reopened tickets remains essential. Where a team is considering automation, it should also check that satisfaction is measured separately for automated and human conversations, so that an improvement in one is not hidden by the other. Pricing and capacity considerations for an AI agent are described on the Bund AI pricing page.
Frequently asked questions
What are the main customer satisfaction metrics? The three most common are CSAT, which measures satisfaction with a specific interaction, NPS, which measures likelihood to recommend the company, and customer effort score, which measures how easy it was to get an issue resolved. Each answers a different question and is collected at a different point.
What is the difference between CSAT and NPS? CSAT is transactional and asks about a recent interaction, usually on a five-point scale. NPS is relational and asks how likely the customer is to recommend the company on a scale of 0 to 10. CSAT suits evaluating support quality, and NPS suits tracking loyalty over time.
How do you calculate NPS and CSAT? NPS equals the percentage of promoters (scores 9 and 10) minus the percentage of detractors (scores 0 to 6). CSAT equals satisfied responses (ratings of 4 and 5) divided by total responses, multiplied by 100. A net promoter score calculator and a CSAT score calculator handle both.
Which customer experience KPIs should a small team start with? Start with CSAT after each closed conversation, first response time and resolution rate. Add customer effort score once a specific process has been identified for improvement, and NPS once volume is high enough for the trend to be reliable.
How often should satisfaction metrics be collected? Collect CSAT after every closed conversation or on a consistent sample, CES after defined transactions, and NPS once or twice a year. Avoid sending several surveys to the same customer after a single interaction.
What is a good customer service KPI for measuring quality? No single customer service KPI captures quality. A reasonable set combines a perception measure such as CSAT, an operational measure such as resolution rate, and an outcome measure such as retention, each reviewed on a fixed cadence.