Articles9 min read
Beyond NPS: what the score misses, and what to ask instead
Net Promoter Score asks one question — “how likely are you to recommend us?” — and produces one number. That number is genuinely useful as a trend line and genuinely useless as a diagnosis: it tells you sentiment moved, never why. This article is about what NPS cannot see, the metrics that complement it, and the follow-up questions that turn a score into a to-do list.
What NPS gets right
Credit first. NPS is short enough that people actually answer it, simple enough that a whole company can rally around it, and — asked the same way at the same cadence — a legitimate directional signal. When the score drops six points after a pricing change, something real happened. The mistake isn’t using NPS; it’s expecting the score to carry information it structurally doesn’t contain.
The four things the score can't tell you
- Why. A 6 from someone frustrated by pricing and a 6 from someone missing a feature are identical in the data and demand opposite responses. Without the reason, teams fill the gap with their favorite theory.
- Which moment. “Recommend us” asks about the whole relationship. It cannot localize the problem to checkout, onboarding, support, or the product itself — the places a fix would actually be applied.
- Who didn’t answer. Single-digit-to-teens response rates mean the score describes your most engaged customers. The quietly-drifting-away middle — the churn risk — is precisely who the survey misses. (More on that skew in survey response rates.)
- Whether anything changed. With modest samples, NPS moves several points on noise alone — a different mix of respondents, a seasonal mood. Celebrating or panicking over small swings is a quarterly ritual at too many companies.
Metrics that fill the gaps
CSAT — satisfaction with a moment
“How satisfied were you with [this interaction]?” asked right after the interaction. Localized where NPS is global: per-touchpoint scores you can assign to an owner.
CES — how hard was it?
Customer Effort Score asks how much work something took. Especially predictive for support and self-serve flows: effort correlates with churn better than delight in much of the CX literature.
PMF survey — how disappointed without it?
“How would you feel if you could no longer use this product?” The Sean Ellis question measures dependence rather than politeness, which makes it sharper than NPS for early-stage products.
Retention — the behavioral truth
What customers do outranks what they say they’d recommend. Renewal, repeat purchase and active use are the scoreboard; survey metrics are leading indicators to be checked against it.
The follow-up is the program
Every one of those metrics shares NPS’s limitation: a number with no reason attached. Which is why the highest-leverage change to any score program costs one question — an open follow-up, matched to the score just given:
- “0–6: What went wrong — and what should we fix first?”
- “7–8: What would move you to a 9 or a 10?”
- “9–10: What would you tell a colleague about us?”
Then treat the verbatims as the product: theme them, count them, and report “the score fell six points and 60% of detractor comments name the new billing flow” instead of the score alone. That sentence is a to-do list; the number never was.
Make the follow-up a conversation
A text box under the score gets one sentence. A conversational follow-up — where an AI interviewer asks about their score, then digs one level deeper — gets the story. That’s how a Feedback Mango study works: tap a rating, then a short interview about why, with every conversation feeding the themed analysis. The mechanics are in AI customer interviews.
Frequently asked questions
- Is NPS still worth using?
- As a lightweight trend line, yes — asked consistently, the score tells you which direction sentiment is moving and gives teams a shared number. As a diagnostic it was never enough: the score contains no reasons, so treat NPS as the smoke detector and pair it with an open follow-up that finds the fire.
- What are the main alternatives to NPS?
- The common companions and alternatives: CSAT (satisfaction with a specific interaction), CES (how much effort something took), PMF surveys ('how disappointed would you be if this disappeared?'), retention and repeat-purchase data as the behavioral truth, and open-ended feedback — interviews or conversational surveys — for the reasons behind any score.
- What should you ask after the NPS question?
- Ask about the score they just gave, matched to the answer: detractors get 'what went wrong — and what should we fix first?', passives get 'what would move you to a 9 or 10?', promoters get 'what would you tell a colleague about us?'. The follow-up answers are where nearly all the actionable value of an NPS program lives.
- Why is my NPS not improving?
- Usually because the score is being tracked but the reasons aren't: without themed follow-up answers, teams guess at causes and fix the wrong things. Less commonly the sample is shifting (different segments respond each wave), which moves the number without anything changing. Fix both by pairing every score with a why and by comparing like segments across waves.
Try it on your site
Let a tiny AI interviewer ask the why
Feedback Mango puts a small AI interviewer on your website: you write one question, it has a short chat with every customer who answers, and the conversations arrive sorted into themes and next steps. Setup is one line of code, and it’s free while in beta — no card.