---
title: "CSAT vs NPS vs CES: Which Customer Metric to Use When"
date: "2026-07-20"
description: "CSAT vs NPS vs CES comes down to three different questions: CSAT measures satisfaction with a specific interaction, NPS measures loyalty and likelihood to recommend the overall relationship, and CES measures how much effort a task took."
keywords: ["csat vs nps vs ces", "csat vs nps", "nps vs ces", "customer experience metrics comparison"]
author: "Perspective AI Team"
category: "Customer Success & Churn Prevention"
slug: "csat-vs-nps-vs-ces-which-customer-metric-to-use-when"
excerpt: "CSAT vs NPS vs CES comes down to three different questions: CSAT measures satisfaction with a specific interaction, NPS measures loyalty and likelihood to…"
image: "https://getperspective.agency/assets/4ebcd0f2-dc8f-49ad-a46d-4b71a04d84a6"
tags: ["customer research", "product management", "csat vs nps", "csat vs nps vs ces", "how-to", "guides"]
lastModified: "2026-07-20"
definition: "CSAT vs NPS vs CES comes down to three different questions: CSAT measures satisfaction with a specific interaction, NPS measures loyalty and likelihood to recommend the overall relationship, and CES measures how much effort a task took. Use CSAT to grade a single touchpoint (a support chat, an onboarding call), NPS to track the health of the whole relationship over time, and CES to find friction in a specific workflow. No single score is \"best\" — they answer different questions and mislead in different ways. The gap they all share is context: Bain & Company found that 80% of companies believe they deliver a superior experience, while only 8% of their customers agree (Bain & Company, \"Closing the Delivery Gap,\" 2005). A number tells you that customers are unhappy; only the \"why\" tells you what to fix."
faqs: [{"question": "What is the difference between CSAT, NPS, and CES?", "answer": "CSAT, NPS, and CES differ by what they measure and at what altitude. CSAT measures satisfaction with a specific interaction on a 1–5 scale, NPS measures overall loyalty and likelihood to recommend on a 0–10 scale, and CES measures how much effort a task required on a 1–7 scale. CSAT and CES are touchpoint-level and captured right after an event, while NPS is a relationship-level metric tracked over time."}, {"question": "Which is better, CSAT, NPS, or CES?", "answer": "No single metric is better; the right one depends on the question you are answering. Use CSAT to grade a specific touchpoint, NPS to track the overall relationship, and CES to find friction in a task. For predicting loyalty after a service interaction, research from the Corporate Executive Board (now Gartner) found CES to be a stronger predictor than CSAT or NPS, but that advantage is specific to service effort, not to every use case."}, {"question": "Can you use CSAT, NPS, and CES together?", "answer": "Yes, and most mature customer experience programs do. A common pattern is to fire CES right after onboarding or issue resolution, CSAT immediately after a specific interaction, and NPS on a quarterly relationship cadence. Using all three at their correct altitude gives you a moment-level, task-level, and relationship-level view without forcing one metric to answer questions it was never designed for."}, {"question": "Why do CSAT, NPS, and CES scores sometimes contradict each other?", "answer": "They contradict because each captures a different slice of experience at a different time. A customer can rate a support chat 5/5 on CSAT (a good interaction) while giving a low NPS (unhappy with pricing or the overall relationship), or report low effort on CES for a task they didn't actually value. Contradictions are a feature, not a bug — they flag where to dig deeper, which requires qualitative follow-up to resolve."}, {"question": "How often should you measure NPS versus CSAT and CES?", "answer": "Measure CSAT and CES event-triggered, immediately after the relevant interaction, and NPS on a periodic cadence, typically quarterly or semi-annually. CSAT and CES are only meaningful while the experience is fresh, so they should be tied to a touchpoint rather than a calendar. NPS is a slow-moving relationship signal, so measuring it too often produces noise and survey fatigue without adding insight."}]
---

## TL;DR

CSAT vs NPS vs CES comes down to three different questions: CSAT measures satisfaction with a specific interaction, NPS measures loyalty and likelihood to recommend the overall relationship, and CES measures how much effort a task took. Use CSAT to grade a single touchpoint (a support chat, an onboarding call), NPS to track the health of the whole relationship over time, and CES to find friction in a specific workflow. No single score is "best" — they answer different questions and mislead in different ways. The gap they all share is context: Bain & Company found that 80% of companies believe they deliver a superior experience, while only 8% of their customers agree ([Bain & Company, "Closing the Delivery Gap," 2005](https://media.bain.com/bainweb/PDFs/cms/hotTopics/closingdeliverygap.pdf)). A number tells you *that* customers are unhappy; only the "why" tells you what to fix.

## What CSAT measures

CSAT (Customer Satisfaction Score) measures how satisfied a customer was with a specific interaction, product, or moment, usually captured immediately after that event. You ask a question like "How satisfied were you with [today's support chat]?" on a 1–5 scale, then report the percentage of respondents who chose the top one or two boxes (a 4 or 5). A 78% CSAT means 78% of respondents were "satisfied" or "very satisfied."

CSAT is the most granular of the three metrics because it is tied to a single touchpoint. That makes it excellent for diagnosing where an experience broke down — a checkout flow, a delivery, a first-response time — and useless for judging whether a customer will stay. Satisfaction is momentary and prone to recency bias: a customer can rate one chat a 5 and churn next month. For the full formula, benchmark ranges, and where the score breaks down, see our pillar on [Customer Satisfaction Score (CSAT)](/blog/customer-satisfaction-score-csat-formula-benchmarks-and-limits). And because "satisfied" is a low bar that hides a lot, it helps to understand [what customer satisfaction actually is](/blog/what-is-customer-satisfaction-how-to-measure-it-beyond-the-score) beyond the top-box number.

## What NPS measures

NPS (Net Promoter Score) measures the strength of the overall customer relationship by asking one question: "How likely are you to recommend us to a friend or colleague?" on a 0–10 scale. Respondents are grouped into promoters (9–10), passives (7–8), and detractors (0–6), and the score is the percentage of promoters minus the percentage of detractors — producing a single number from −100 to +100.

NPS is a relationship-level, not touchpoint-level, metric. It is designed to be tracked over time as a directional signal of loyalty and word-of-mouth growth, which is why it sits at the center of most executive dashboards. Its weakness is that one number compresses hundreds of different reasons into a single digit: two companies with an NPS of 40 can have completely different problems. To use it well, you need the definition and the reasoning behind it — start with our guide to [what Net Promoter Score (NPS) is](/blog/what-is-net-promoter-score-nps-definition-formula-and-the-why-behind-the-score), then get precise on the buckets with [the NPS scale explained](/blog/the-nps-scale-explained-0-10-promoters-passives-and-detractors). NPS matters commercially because loyalty compounds: Bain & Company research popularized in Harvard Business Review shows that a 5% increase in customer retention can lift profits by 25% to 95% ([Harvard Business Review, "The Value of Keeping the Right Customers," 2014](https://hbr.org/2014/10/the-value-of-keeping-the-right-customers)).

## What CES measures

CES (Customer Effort Score) measures how much effort a customer had to expend to get something done — resolve an issue, complete a purchase, finish onboarding. The modern version asks respondents to agree or disagree with a statement like "The company made it easy for me to handle my issue" on a 1–7 scale, then averages the responses. Lower effort is better.

CES is the youngest of the three metrics and the most predictive of future behavior for service interactions. It was introduced by the Corporate Executive Board (now Gartner) in the 2010 Harvard Business Review article ["Stop Trying to Delight Your Customers,"](https://hbr.org/2010/07/stop-trying-to-delight-your-customers) which studied more than 75,000 customers and found that reducing effort predicts loyalty better than trying to exceed expectations. In that research, CES was a stronger predictor of loyalty than either customer satisfaction or Net Promoter Score, and 96% of customers who had a high-effort service experience became more disloyal, versus only 9% of those with a low-effort experience. CES's blind spot is the mirror image of its strength: it only sees friction. A frictionless experience can still be one nobody values, so CES tells you where you are annoying customers but not whether you are delivering something worth their time.

## CSAT vs NPS vs CES: side-by-side comparison

The fastest way to keep CSAT vs NPS vs CES straight is to line them up by the question they ask, the scale they use, and the blind spot they carry. Each metric is optimized for a different altitude — a single moment, the whole relationship, or a specific task — so they are complements, not substitutes.

| Dimension | CSAT | NPS | CES |
|---|---|---|---|
| Core question | "How satisfied were you with [X]?" | "How likely are you to recommend us?" | "How easy was it to [do X]?" |
| Typical scale | 1–5 (top-box) | 0–10 | 1–7 (agree/disagree) |
| Reported as | % who chose 4–5 (0–100%) | Promoters % − Detractors % (−100 to +100) | Average score (or % low-effort) |
| Altitude | Single touchpoint or moment | Overall relationship | A specific task or workflow |
| Best timing | Immediately after an interaction | Periodic, relationship-level | Immediately after an effortful task |
| Strongest at predicting | Whether that touchpoint worked | Advocacy and long-term loyalty (directional) | Loyalty and churn for service issues |
| Main blind spot | "Satisfied" ≠ loyal; recency bias | One number hides the reason | Sees friction, not value or delight |

If you want the wider dashboard these three sit inside — including retention, lifetime value, and health scores — our overview of [customer experience metrics: the 8 that matter](/blog/customer-experience-metrics-in-2026-the-8-that-matter-nps-csat-ces-clv-and-more) maps how they fit together, and [customer service metrics: 12 KPIs that matter](/blog/customer-service-metrics-12-kpis-that-matter-and-what-they-miss) covers the operational layer beneath them.

## Which metric to use when: a decision framework

Choose the metric by matching its altitude to the question you are trying to answer, not by picking a company-wide favorite. The three map cleanly onto three different jobs, and most mature programs run all three at different moments in the journey rather than forcing one to do everything.

Use this quick decision guide:

- **Use CSAT when** you want to grade a specific, recent touchpoint — a support ticket, a delivery, a demo, an onboarding session. Fire it immediately after the event while memory is fresh. It answers: *did this particular interaction work?*
- **Use NPS when** you want a periodic read on the overall relationship and appetite to advocate — quarterly relationship surveys, post-renewal check-ins, executive dashboards. It answers: *how healthy is the relationship over time?*
- **Use CES when** you want to find and remove friction in a task the customer had to complete — resolving an issue, finishing setup, checking out. Fire it right after the effortful step. It answers: *where are we making this harder than it needs to be?*

A practical way to combine them across the lifecycle:

1. **Onboarding:** CES right after setup ("How easy was it to get started?") to catch friction before it becomes churn.
2. **Support:** CSAT immediately after ticket resolution, plus CES on the same interaction to separate "were they friendly" from "was it easy."
3. **Product/UX:** CSAT on a specific feature or release to see if a change landed.
4. **Relationship:** NPS on a quarterly or semi-annual cadence to track loyalty direction and flag accounts trending toward detraction.

A common mistake is treating a high score as proof you are safe. Satisfaction and effort are both moment-bound, and NPS is directional at best. Remember the delivery gap: executives systematically overrate their own experience, so a green dashboard is a hypothesis, not a conclusion. Before you standardize on question wording and cadence, it helps to see real examples — our library of [CSAT survey questions: examples and templates for 2026](/blog/csat-survey-questions-examples-and-templates-2026) gives you copy-ready wording, and if the goal is to move the number rather than just watch it, start with [how to improve customer satisfaction](/blog/how-to-improve-customer-satisfaction).

## Why all three metrics need qualitative context

Every one of these metrics is a symptom tracker, not a diagnosis — the score tells you something changed, but never why. A CSAT of 62% does not tell you which part of the checkout frustrated people. An NPS drop from 45 to 30 does not tell you whether it was pricing, a bug, or a support failure. A high CES on a task nobody values still leaves you blind to the real problem. This is the delivery gap in action: leaders read scores without reading customers.

The traditional fix — a free-text "Why did you give that score?" box appended to the survey — mostly fails. Most respondents skip it, and those who answer write a few flat words ("too slow," "fine") that a static form can never probe. The highest-value context lives in the answers people don't volunteer: the "it depends," the workaround they built, the reason they almost left. Forms flatten those into dropdowns and blank fields.

This is where Perspective AI changes the shape of the problem. Instead of bolting a comment box onto a score, Perspective AI runs the follow-up as a real conversation: an [AI interviewer agent](/agents/interviewer) asks a detractor *why* they scored a 3, follows up on a vague answer, and probes the constraint behind the complaint — at the scale of hundreds or thousands of respondents at once. For teams that own these metrics, this means the number and its explanation arrive together. It is purpose-built for [CX teams](/roles/cx-teams) drowning in scores they can't act on and for [product teams](/roles/product-teams) trying to turn a CSAT dip into a roadmap decision. You can even replace the survey form itself with a [conversational concierge agent](/agents/concierge) so the "why" is captured in the moment rather than chased down later.

## Frequently Asked Questions

### What is the difference between CSAT, NPS, and CES?

CSAT, NPS, and CES differ by what they measure and at what altitude. CSAT measures satisfaction with a specific interaction on a 1–5 scale, NPS measures overall loyalty and likelihood to recommend on a 0–10 scale, and CES measures how much effort a task required on a 1–7 scale. CSAT and CES are touchpoint-level and captured right after an event, while NPS is a relationship-level metric tracked over time.

### Which is better, CSAT, NPS, or CES?

No single metric is better; the right one depends on the question you are answering. Use CSAT to grade a specific touchpoint, NPS to track the overall relationship, and CES to find friction in a task. For predicting loyalty after a service interaction, research from the Corporate Executive Board (now Gartner) found CES to be a stronger predictor than CSAT or NPS, but that advantage is specific to service effort, not to every use case.

### Can you use CSAT, NPS, and CES together?

Yes, and most mature customer experience programs do. A common pattern is to fire CES right after onboarding or issue resolution, CSAT immediately after a specific interaction, and NPS on a quarterly relationship cadence. Using all three at their correct altitude gives you a moment-level, task-level, and relationship-level view without forcing one metric to answer questions it was never designed for.

### Why do CSAT, NPS, and CES scores sometimes contradict each other?

They contradict because each captures a different slice of experience at a different time. A customer can rate a support chat 5/5 on CSAT (a good interaction) while giving a low NPS (unhappy with pricing or the overall relationship), or report low effort on CES for a task they didn't actually value. Contradictions are a feature, not a bug — they flag where to dig deeper, which requires qualitative follow-up to resolve.

### How often should you measure NPS versus CSAT and CES?

Measure CSAT and CES event-triggered, immediately after the relevant interaction, and NPS on a periodic cadence, typically quarterly or semi-annually. CSAT and CES are only meaningful while the experience is fresh, so they should be tied to a touchpoint rather than a calendar. NPS is a slow-moving relationship signal, so measuring it too often produces noise and survey fatigue without adding insight.

## Conclusion: pick the metric, then chase the why

The CSAT vs NPS vs CES decision is not about crowning a winner — it is about matching the metric to the altitude of your question. Reach for CSAT to grade a specific touchpoint, NPS to track the overall relationship, and CES to hunt down friction in a task. Run all three where they fit, and stop expecting any one of them to explain itself.

What every score has in common is that it is a starting point, not an answer. The 80%-versus-8% delivery gap Bain documented survives because teams optimize numbers they don't truly understand. The metric tells you *that* something moved; the customer's own words tell you *what to do about it*. That is the layer Perspective AI adds — conversational interviews that capture the reasoning behind every score, at survey scale. [Start a research study](/research/new) to attach the "why" to your next CSAT, NPS, or CES reading, [browse example studies](/studies) to see the format in action, or [compare the approach](/compare) and check [pricing](/pricing) when you are ready to move beyond the score.