CSAT vs NPS vs CES: Which Customer Metric Should You Actually Use?

Three customer-feedback metrics dominate the conversation. They're not interchangeable. Most teams either pick one and over-rely on it, or pick all three and run them at the same touchpoints — which produces overlapping data and survey fatigue.

Here's exactly when to use each, with concrete examples.

The three metrics in one sentence each

• CSAT (Customer Satisfaction Score): how satisfied are you with this specific interaction?
• NPS (Net Promoter Score): how likely are you to recommend us overall?
• CES (Customer Effort Score): how much effort did it take to do the thing you just did?

Each measures something different. CSAT is about a specific touchpoint. NPS is about the relationship. CES is about friction at a specific moment.

CSAT in detail

The question: "Overall, how satisfied are you with [the specific interaction]?" — usually 1–5 scale.

The math: Most teams report CSAT as the percentage of respondents who give 4 or 5 ("satisfied" or "very satisfied"). So CSAT 85% means 85% of respondents rated the interaction 4 or 5 out of 5.

When to use:

• Right after a support ticket closes ("How satisfied were you with how we resolved your issue?")
• Right after an order delivers ("How satisfied are you with your order?")
• Right after onboarding completes ("How satisfied are you with your setup experience?")
• Right after a specific feature is used ("How satisfied are you with [feature]?")

The pattern: CSAT works when there's a clear, recent, definable interaction the customer can rate.

When NOT to use:

• For overall customer relationship sentiment (that's NPS)
• For the underlying difficulty of an interaction (that's CES)
• Months after an event (memory has faded; use NPS instead)

See our CSAT template →

NPS in detail

The question: "How likely are you to recommend us to a friend or colleague?" — 0–10 scale.

The math: Promoters (9–10) minus Detractors (0–6), as a percentage. Range -100 to +100.

When to use:

• Quarterly check-in across your customer base
• Embedded in your product as ambient sentiment
• After major product/relationship milestones (annual contract renewal, major release)

The pattern: NPS measures the durable relationship, not a specific interaction. It's the "would you bet on us?" question.

When NOT to use:

• For specific touchpoints (use CSAT)
• For ease-of-use measurement (use CES)
• Too frequently — quarterly is the sweet spot; weekly creates fatigue and noise

See our NPS template → or read the full NPS guide →.

CES in detail

The question: "How easy was it to [complete the action]?" — usually 1–5 or 1–7 scale.

Standard version (from Gartner research): "[Company name] made it easy for me to handle my issue." Scale: Strongly Disagree → Strongly Agree.

The math: Average score on the scale, or % of respondents who agree (4–5 on a 5-point scale, 6–7 on a 7-point scale).

When to use:

• After self-serve actions (signup, configuration, integration setup)
• After support interactions (especially complex ones)
• For onboarding flows where friction is the bottleneck
• For the product's most-used workflows

The pattern: CES measures friction. Use it when the workflow is the unit of analysis, not the company or the support ticket.

When NOT to use:

• For emotional satisfaction (use CSAT)
• For long-term loyalty (use NPS)
• For workflows that aren't supposed to be effortless (e.g., complex financial decisions)

The most common mistake: bundling all three

Lots of teams do this — they run an NPS survey that's actually 8 questions: an NPS question, three CSAT-style ratings ("rate our product, our support, our pricing"), a CES question, and three open-text questions.

This produces useless data for all three metrics:

• The NPS number can't be compared to industry benchmarks because the survey context isn't standardized
• The CSAT scores within an NPS survey are biased by overall sentiment
• The CES score in the middle of an NPS survey doesn't tie to any specific recent action
• The respondent is fatigued by question 5 and click-throughs the back half

The fix: keep each metric to its own focused survey, sent at the right trigger.

Metric	Survey	Cadence	Length
CSAT	After specific events	Triggered, immediate	3 questions
NPS	Quarterly to whole base	Calendar-based	2 questions
CES	After specific workflows	Triggered, immediate	3 questions

Each survey is short. Each lives at its own touchpoint. None overlap.

When you only have time/budget for one

If you can only invest in one customer feedback program, choose based on your stage:

Early stage (pre-product-market-fit): Choose CES. You need to measure where friction is in your core workflows. The unit of analysis is the workflow, not the relationship — and you don't have a stable relationship yet.

Growth stage (post-PMF, scaling): Choose NPS. You need to know whether your relationship-level sentiment is healthy as you scale. The leading indicator of retention.

Mature stage (steady-state): Choose CSAT. The relationship is established; you need to keep specific touchpoints working. Catch friction at the moment before it compounds into churn.

Most teams default to NPS because it's the most quoted metric in board decks. NPS is the right answer for some stages and not others.

Combining all three (when you actually have resources)

The mature customer-feedback program uses all three at different touchpoints:

1. NPS quarterly to entire customer base (broad health metric for the board)
2. CSAT after every support ticket (5%-segment of tickets is fine, don't need 100%)
3. CSAT after onboarding (every new account)
4. CES after major workflow events (signup, integration setup, first import)

Each metric answers a different question. The dashboard shows all three trending over time, plus the text-comment patterns from each.

The key: never put two metrics in the same survey. Each gets its own focused 2-4 question form, triggered at the right moment.

How to interpret the numbers

Wildly varies by industry. Use these rough ranges:

CSAT (% scoring 4 or 5):

• B2B SaaS support tickets: 85%+ is good; 90%+ is exceptional
• Consumer e-commerce orders: 90%+ is good
• Onboarding flows: 70-85% is normal; below 60% suggests friction issues

NPS (industry varies wildly, see NPS guide):

• B2B SaaS: 30-50 is healthy
• Consumer SaaS: 20-40
• Telecom: often negative

CES (% Agree/Strongly Agree):

• Self-serve signup flows: 70%+ is good; below 60% means friction
• Support: 70%+; complex tier-2 tickets often 50-65%
• Common workflows in mature products: 85%+

Don't chase absolute numbers. Watch your own trend over 4+ quarters.

Bottom line

CSAT for specific interactions. NPS for the relationship. CES for friction.

Run each at the right trigger, in its own focused survey, at the right cadence. Don't bundle. Don't over-collect. Read the open-text responses (where the value is).

Start a CSAT survey → — or NPS → — or browse all 30 templates →.