Most SaaS dashboards stack all three side by side. The result: none of them triggers any action, because you're always waiting for the three to align. They never do.
CSAT: what does it measure, and when should you use it?
- Measures: "Are you satisfied with this interaction?" on a 5-point scale, right after the conversation.
- Sample: respondents only, self-selection bias owned.
- Cadence: weekly review per team.
- Decisions it triggers: agent coaching, macro tuning, quality escalation.
CES: what does it measure, and when should you use it?
- Measures: "How much effort did it take to resolve?" on a 7-point scale.
- Strength: the best known predictor of retention. Correlates better than CSAT and NPS.
- Weakness: wording-sensitive. The trend matters more than the absolute value.
- Decisions it triggers: product roadmap, self-serve, journey redesign.
NPS: what does it measure, and when should you use it?
- Measures: "Would you recommend?" on 0-10.
- Strength: board can grasp it in 8 seconds.
- Weakness: noisy in B2B, can swing on 5 detractors out of 50 respondents, sensitive to survey timing.
- Decisions it triggers: very few internal ones. Good for external comms.
Recommended setupWeekly: CSAT per team. Monthly: CES trend per persona. Quarterly: NPS for the board. Three cadences, three uses, zero confusion.
Which mistakes should you avoid with CSAT, CES and NPS?
- Target an absolute NPS. Targets must be relative to baseline.
- Over-index on negative CSAT comments, they're over-represented by construction.
- Mix support CSAT and product CSAT in the same dashboard. You're hiding the signal.
A metric that doesn't trigger any action within 7 days is a cost, not a steering tool.