NPS, CSAT or CES: which metric should you choose to measure customer satisfaction?

Why measure customer satisfaction with a dedicated metric

Customer satisfaction cannot be managed by intuition alone. Yet that is still the reality for the majority of SMBs: you know a client is happy because they call back, or dissatisfied because they started talking to a competitor. This reactive approach has a direct — and measurable — cost.

B2B retention research shows that a 5% increase in the customer retention rate can increase profitability by 25% to 95%. But to retain, you first need to understand — and to understand, you need to measure in a structured way.

Three metrics dominate the customer satisfaction landscape today: NPS (Net Promoter Score), CSAT (Customer Satisfaction Score) and CES (Customer Effort Score). Each measures something fundamentally different. Using the wrong one in the wrong context means making decisions based on false premises.

NPS — measuring the propensity to recommend

The Net Promoter Score is the most widely used customer satisfaction metric in the world. Created by Fred Reichheld in 2003 and published in the Harvard Business Review, it is based on a single direct question.

When to use NPS

NPS is a relational indicator. It measures overall loyalty towards your company — not a specific transaction. It is particularly relevant:

• On a quarterly or semi-annual cadence, across your entire customer base
• As an overall health indicator to track over time (trend matters more than the absolute score)
• As a reference metric in your leadership reports and dashboards
• To identify at-risk customers (detractors) and potential brand ambassadors (promoters)

Strengths and limitations

CSAT — measuring transactional satisfaction

The Customer Satisfaction Score is the most intuitive of the three metrics: it directly asks whether the customer was satisfied with a specific interaction. Unlike NPS, which measures overall loyalty, CSAT captures satisfaction in the moment, right after a triggering event.

When to use CSAT

CSAT is a transactional indicator. It excels in contexts where you want to evaluate a specific moment in the customer experience:

• After a customer service call or support interaction
• After the delivery of a project or the implementation of a solution
• After a product update or the launch of a new feature
• At each key touchpoint in your customer journey (onboarding, renewal)

Strengths and limitations

CES — measuring friction in the experience

The Customer Effort Score is the least well-known of the three — and yet, according to multiple studies from Gartner and the Customer Contact Council, it is one of the strongest predictors of customer loyalty. Its premise is counter-intuitive: customers are not looking to be delighted — they are looking to resolve their problem as easily as possible.

When to use CES

CES is ideal for measuring the operational fluidity of your service:

• After a support interaction (ticket resolution, refund request)
• During or after the onboarding process for a new customer
• After a complex administrative process (billing, contract update)
• To evaluate the ease of use of your product or platform

Strengths and limitations

Direct comparison: NPS vs CSAT vs CES

Here is a summary view of the three metrics to guide your decision:

How to choose the right metric for your SMB

The honest answer: start with a single metric, well defined, well measured, well acted upon. An NPS collected every quarter with systematic follow-ups is worth infinitely more than a three-metric dashboard that nobody knows how to respond to.

If you are starting a feedback programme

Start with NPS. It is simple to explain to your team, has industry benchmarks (so you know where you stand), and the associated open-ended question ("Why this score?") generates the richest verbatims to guide your priorities.

If you already have an NPS programme in place

Add CSAT at your critical touchpoints: onboarding, renewal, support. You will move from an annual view to a dynamic one — seeing in real time which interactions are degrading (or strengthening) the overall relationship.

If your main challenge is retention and churn

Integrate CES into your support process. A low CES after a support interaction is the most reliable signal of a short-term departure risk. It is an early warning that neither NPS nor CSAT detects as early.

Reference benchmarks — where do you stand?

Benchmarks vary significantly by sector. Here are the reference points for SaaS and B2B service SMBs:

Important: benchmarks are reference points, not absolute targets. What matters more is your progression over time. An NPS that moves from 18 to 34 over two years, in a sector where the median is 28, is a far more compelling traction signal for an investor than a score of 50 that has been flat for three years.

For more on NPS benchmarks by sector, see our article on how to calculate your NPS and interpret industry benchmarks.

Common mistakes to avoid

Collecting satisfaction metrics is straightforward to set up — and easy to get wrong. Here are the most frequent pitfalls in SMBs.

1. Measuring without acting. A feedback programme that generates no visible action for customers damages the relationship. If you collect, you must show that you listen — even a simple response to a detractor changes perception.
2. Sending too often. Survey fatigue is real. A customer solicited after every micro-interaction will eventually ignore your messages — or worse, respond carelessly. Calibrate the cadence based on the type of metric and the nature of the relationship.
3. Treating the score as a vanity KPI. An NPS of 45 has no value if nobody in your organization is accountable for improving it. The metric must be tied to concrete actions and designated owners.
4. Neglecting segmentation. An overall NPS of 35 can hide an NPS of 55 among your enterprise accounts and an NPS of 12 among your recent SMB clients. Segmentation reveals where to act.
5. Ignoring verbatims. Scores tell you how many. Verbatims tell you why. Analysing open-ended responses — especially at scale — is where real business intelligence is generated.