What Is a Good NPS Score? How to Read Yours (Without Fake Benchmarks

What is a good NPS score? How to read yours (without fake benchmarks)

You ran your first NPS survey, and the number came back as +18. Or +42. Or −3. Now you want to know: is that good?

The short answer is a set of general bands that the industry uses as a rough guide. The honest answer is that “good” depends almost entirely on your own previous score, not on a table of industry averages you found online. This post gives you both — the bands and the reason to ignore most benchmark tables — so you know exactly how to read your number and what to do with it.

If you need the formula first, see [NPS vs CSAT vs CES: which metric should you track?][PILLAR: customer-feedback-metrics-nps-csat-ces] or run your numbers through the NPS calculator.

The general NPS bands

NPS runs from −100 (every respondent is a detractor) to +100 (every respondent is a promoter). The commonly used interpretation bands are:

[table]

These bands are widely used in the NPS community and cited by practitioners including Bain & Company, the firm behind the original Net Promoter system. They’re a useful sanity check — if your score is negative, something is clearly wrong — but they were never meant to be the final word on whether your programme is working.

The number that matters most is your own previous score. An NPS of +18 that used to be +5 is a business getting better. An NPS of +55 that used to be +68 is a business losing loyalty. The trend tells you more than the absolute number.

Why “average NPS by industry” tables are mostly noise

Search for “NPS benchmark by industry” and you’ll find dozens of tables listing precise averages — “SaaS: +41”, “Retail: +54”, “Healthcare: +38”. Most of them have three problems:

1. You don’t know who was surveyed. Was it a broad consumer panel, or a company’s own paying customers? The two groups give very different scores. A panel of non-customers rates brands lower because they have no relationship to evaluate.

2. You don’t know when or how. A relational NPS survey sent quarterly gives a different score than a transactional one sent after a purchase. If the benchmark table doesn’t tell you the method, the number is meaningless.

3. The sample may be tiny or biased. Some widely cited “industry averages” come from a few hundred responses across an entire sector. Others come from one research firm’s client base, which skews toward companies that care enough about CX to hire a research firm — not exactly a representative sample.

What to do instead: Track your own score over time, with the same method, the same question, the same timing, and the same audience. When NPS moves from +22 to +35 across two quarters, you know something real changed. When it drops from +35 to +28, you know to look at the open-text follow-ups and find out why.

If you want an external reference point, the broadest public data comes from SurveyMonkey’s 2025 global benchmark report, which puts the global median at around +44 based on 150,000+ organisations. But even that number varies dramatically by geography, company size, and method — so treat it as a rough compass, not a target.

What actually moves an NPS score

Before you panic about a low number or celebrate a high one, consider the five things that move NPS independently of how good your business actually is:

1. Who you survey. Survey only your best customers and the score soars. Survey everyone, including people who signed up and never came back, and it sinks. Neither is wrong, but they’re not comparable.

2. When you survey. Right after a purchase (when excitement is high) produces higher scores than three months later (when the product has had time to disappoint). Relational and transactional surveys produce different baselines — more on that below.

3. How you send it. A WhatsApp message with a direct link gets faster, more emotionally fresh responses than an email that sits in a crowded inbox for three days. The channel changes the score.

4. How often you survey. Over-surveying fatigues respondents. The ones who keep answering tend to be either very loyal or very frustrated — the middle drops out, and your score gets more extreme.

5. Response rate. A 5% response rate means you’re hearing from the edges — people who love you or hate you. A 30% rate gives you the passives too, which anchors the score closer to reality. See [how to improve your survey response rate][SPOKE: how-to-improve-survey-response-rate].

This is why comparing your NPS to another company’s is unreliable unless you know they used the same method, timing, audience, and channel. And you almost never know that.

How to read a change in your score

A real NPS change is one where the method stayed the same and only the customer experience changed. Here’s how to read it:

A jump of 5–10 points in one quarter is meaningful — something changed. Check the open-text follow-ups from promoters (what did you start doing right?) and detractors (what got worse or better?).

A jump of 15+ points is significant. Either you fixed a major pain point, or something about your method changed (new audience, new channel, different timing). Rule out the method change first.

A steady score across 3–4 quarters means your customer experience is stable. That’s fine if the score is healthy, and a problem if it’s not — stable at +12 means nothing is improving.

A gradual decline over 2+ quarters is the most dangerous pattern, because it’s easy to ignore. Each quarter’s drop is small, but the trend means loyalty is eroding. This is when CES data is most useful — a rising effort score alongside a declining NPS almost always points to a support or process problem. See [Customer Effort Score: the question, scale & when to use it][SPOKE: customer-effort-score-survey].

Worked example: Q1: +22 (200 responses). Q2: +31 (210 responses). Q3: +28 (195 responses). The Q1→Q2 jump is real — 9 points on similar sample sizes. The Q2→Q3 dip is worth watching but could be noise at this volume. If Q4 comes in below +25, the trend is real and it’s time to dig into the follow-ups.

Relational vs transactional baselines

If you run both relational and transactional NPS (and most mature programmes do), expect different baselines:

•  Relational NPS tends to be higher, because you’re asking about the whole relationship and people weigh the positives more when they’re not reacting to a specific moment.

•  Transactional NPS is more volatile. A great support experience can produce a 9 from someone whose relational score is a 7. A frustrating checkout can produce a 3 from someone whose relational score is an 8.

Keep them on separate dashboards. A transactional drop tells you which touchpoint broke. A relational drop tells you the overall relationship is weakening. They’re both useful — they’re just not the same number.

For more on the difference, see [NPS survey questions: the core question, follow-ups & 20 examples][SPOKE: nps-survey-questions].

What to do with your score

The score is not the insight. The insight is in the combination of the score and the follow-up answers. Here’s a simple framework:

If your NPS is below 0: You have more detractors than promoters. Read every detractor follow-up this quarter. Look for the top 2–3 themes. Those are your fires — fix them before you do anything else.

If your NPS is 0–30: You’re in healthy territory, but you have a passive problem — lots of people who are fine but not enthusiastic. Read the passive follow-ups. Look for “it’s okay, but…” answers. The thing after “but” is your opportunity.

If your NPS is above 30: Your promoters are strong. Now protect them — read the promoter follow-ups to understand why they love you, and make sure you don’t change those things by accident. Also check your CES: high NPS with high effort means your promoters are loyal despite friction, which is fragile.

If your NPS is above 70: You’re in rare territory. The risk here is complacency — the score is so good that nobody looks at the follow-ups. Keep reading them. And start watching for the “passive creep” pattern: promoters sliding to passives, passives sliding to detractors, one quarter at a time.

Run your numbers: NPS calculator. Start collecting: NPS survey template.

Frequently Asked Questions related to what is good NPS Score

What NPS score is considered good?

The general bands are: below 0 needs attention, 0–30 is good, 30–70 is great, above 70 is excellent. But the number that matters most is your own previous score — an NPS of +18 that was +5 last quarter is a business improving fast.

What is the average NPS score?

It depends entirely on who was surveyed, how, and when. The broadest public data (SurveyMonkey’s 2025 report, 150,000+ organisations) puts the global median at roughly +44. Your own baseline, measured consistently, is more useful than any external average.

Is an NPS of 50 good?

Yes — +50 falls in the “great” band and means you have significantly more promoters than detractors. But context matters: +50 that used to be +65 is a warning, while +50 that used to be +30 is a success.

Why did my NPS drop even though nothing changed?

Something probably did change, but not always in your product. Common causes: you surveyed a different audience segment, changed the survey channel (email to in-app), sent it at a different point in the journey, or your response rate shifted. Rule out method changes before blaming the product.

Should I compare my NPS to competitors?

Only if you know they use the same method, timing, audience, and channel — and you almost never do. Benchmarking against your own trend is more reliable and more actionable.

Try it

Templates - NPS survey template - SaaS product NPS template - Customer effort score template

Calculators - NPS calculator - CSAT calculator - CES calculator

Product - Survey & feedback software · WhatsApp surveys