NPS vs CSAT: Which Feedback Metric Should You Actually Track?

Most teams pick NPS because it's the metric they've heard of. Then they run it for six months, the score barely moves, and nobody can say what to build differently because of it.

That's not an NPS problem. It's a metric-fit problem. NPS and CSAT answer different questions, and picking the wrong one means collecting numbers you can't act on.

Here's the short version: NPS measures long-term loyalty. CSAT measures the experience someone just had. If you're iterating on a product, CSAT per touchpoint tells you what to ship next. NPS is a board-meeting number — useful, but trailing.

What NPS actually measures

NPS asks one question: "How likely are you to recommend us to a friend or colleague?" on a 0–10 scale. Scores of 9–10 are promoters, 7–8 are passives, 0–6 are detractors. Your score is the percentage of promoters minus the percentage of detractors, landing somewhere between −100 and +100.

Two things follow from that design:

It's a relationship metric, not a feature metric. When someone answers, they're averaging everything — your product, your pricing, that one support ticket that took four days. A low score tells you something is wrong somewhere. It doesn't tell you where.

It's a trailing indicator. Loyalty is the accumulation of many experiences. By the time your NPS drops, the experiences that caused it happened weeks or months ago. You're reading last quarter's mistakes.

There's also a volume problem for small products. The promoters-minus-detractors math amplifies noise: with 50 responses, a handful of annoyed users can swing your score 10 points. If you have a few hundred active users, your NPS will jump around for reasons that have nothing to do with your product.

None of this makes NPS useless. It's a decent health check, it's comparable across companies, and investors recognize it. But it's a thermometer, not a diagnosis.

What CSAT actually measures

CSAT asks "How satisfied were you with this?" — usually on a 1–5 scale, right after a specific experience. Onboarding, an export, a support reply, a new feature. The score is the percentage of respondents who picked 4 or 5.

Because it's anchored to one touchpoint, CSAT gives you something NPS can't: a location. If checkout satisfaction is 92% and your import flow sits at 61%, you know where next sprint goes. No detective work between survey and cause.

It's also a leading indicator. Users tell you the import flow is frustrating this week — not after the frustration has compounded into churn and a detractor score.

The trade-off: CSAT tells you nothing about overall loyalty. Users can be perfectly satisfied with every individual interaction and still leave for a competitor that solves the problem better. Touchpoint satisfaction is necessary, not sufficient.

Where CES fits

Customer Effort Score asks "How easy was it to get this done?" It's CSAT's sibling, tuned for friction: support resolutions, setup flows, anything where the user had a job to finish. If your product's core promise is "this used to be tedious and now it isn't," CES per flow is worth tracking. For most small SaaS teams it's a supplement, not the primary metric.

How to choose

Match the metric to the question you're actually asking:

You want to know	Track	Cadence
What should we fix or build next?	CSAT per touchpoint	Continuous, after the experience
Is this flow too much work?	CES on that flow	Continuous, after completion
Are we building durable loyalty?	NPS	Quarterly, relationship-wide

Stage matters more than industry here:

• Pre-product-market fit, small user base: CSAT on your core flows. You need per-feature signal, and you don't have the response volume for NPS to be stable anyway.
• Growing, retention becoming the question: add quarterly NPS as a health check. The follow-up comments from detractors are usually worth more than the score.
• Reporting to a board: they'll ask for NPS. Give it to them — just don't run your sprint planning off it.

You don't need two tools for this

The practical reason teams stick with one metric is tooling: one widget is installed, and adding a second survey vendor for a different question type is nobody's idea of a good afternoon.

That's the part worth fixing, because it's now a config change rather than an integration. SeggWat's survey widget ships templates for NPS, CSAT, and CES, and each survey has its own trigger rules — so "CSAT after onboarding completes" and "NPS on a 90-day cooldown" run side by side from the same snippet.

Fire the CSAT survey from a custom event when a flow completes, keep NPS on a time-based trigger with a cooldown so nobody gets double-surveyed, and read both in one dashboard. The metric choice stops being lock-in, which means you can pick the honest answer instead of the installed one.

The bottom line

Track CSAT to decide what to build. Track NPS to check whether it's working. If you can only instrument one thing this week and you're still iterating on the product, make it CSAT after your riskiest flow — that's the survey that changes what you ship.

Try SeggWat free — the survey widget installs in one script tag, and the NPS, CSAT, and CES templates are ready to trigger.