What Is a Good NPS Score? Benchmarks, Context and What the Number Really Tells You

A good NPS score, according to Bain & Company who created the metric, is anything above 0. Above 20 is favorable, above 50 is excellent, and above 80 is world class. That is the benchmark almost everyone quotes, and it is a reasonable starting point. It is also close to useless on its own, because a score that looks weak in one industry is strong in another, and because the number tells you nothing about what to do next. Here is the context that makes it usable.

The standard NPS benchmarks

Bain also found that companies achieving long-term profitable growth tend to carry an NPS roughly twice the average for their industry, which is the most defensible version of the "NPS predicts growth" claim.

Why the benchmark alone will mislead you

Industry gravity is real and it is large. Sector averages differ so much that the same raw number can be a triumph or a disaster depending on what you sell. Published industry averages have put groceries around 30 and video streaming around 29, while consumer payments has come in negative. Our breakdown of NPS benchmarks by industry lays out the typical range for SaaS, ecommerce, financial services and telecom side by side. If you are running a product in a structurally disliked category, clearing 0 may be genuinely excellent work. If you are in a category people love, a 35 might mean you are losing.

Then there is the mechanics of the score itself. NPS is a difference of two percentages, not an average, so it moves violently on small samples. A quarterly blast to a short list can swing ten points on a handful of responses and mean absolutely nothing. Before you react to a change, check whether the sample changed.

A B2B SaaS product at NPS 35 with a shrinking, well-understood list of detractor complaints is in far better shape than one at 45 that has no idea why.

How NPS is calculated

One question: how likely are you to recommend us to a friend or colleague, on a scale of 0 to 10.

• Promoters score 9 or 10.
• Passives score 7 or 8.
• Detractors score 0 through 6.

NPS is the percentage of promoters minus the percentage of detractors, giving a range of -100 to +100. Passives are counted in the denominator but contribute to neither side, so a pile of lukewarm 7s and 8s will hold your score down without ever showing up as a complaint. That is worth remembering: passives are not neutral in effect, they are a drag.

Your own trend beats any benchmark

The only NPS comparison that reliably means something is your score against your own score, segmented and tracked over time. Specifically:

• By plan or tier. Enterprise accounts and self-serve users usually have completely different experiences. Averaging them produces a number that describes nobody.
• By customer age. Newly onboarded users and three-year veterans score for different reasons. A drop concentrated in month-one accounts is an onboarding problem, not a product problem.
• By segment or use case. Very often one underserved segment is dragging the whole score, and fixing that one group moves everything.

A blended company-wide NPS is the corporate equivalent of taking the average temperature of a hospital. It is technically a number.

Why is my NPS score dropping?

A falling NPS almost never has a cause you can see in the score. In practice there are three usual suspects, and you cannot tell them apart without looking at the comments.

• The mix of respondents changed. A new, less well served segment started answering. The product did not get worse, your audience got broader.
• Something specific regressed. A pricing change, a redesign, a bug, a support backlog. This shows up as a sharp, thematically concentrated cluster of detractor comments.
• Friction accumulated. The slow one. Passives quietly tip into detractors over months as small annoyances add up, and there is no single event to point at.

The way to tell which you are dealing with is to read what detractors wrote and look at what they did. If the comments cluster on one theme, and that theme grew in the same period the score fell, you have your driver. If the comments are diffuse but the respondent mix shifted, you have a sampling story, not a product one.

The follow-up question is the whole point

The score is a headline. The free-text answer underneath is the article. Almost everything actionable in an NPS program lives in that comment box, and in most companies it is never systematically read, because reading a few thousand open-ended responses by hand is a job nobody has time for.

That constraint is gone. Open-ended responses can now be read in full, clustered into the themes that genuinely repeat, and quantified, so instead of "NPS fell 4 points" you get "NPS fell 4 points, driven by 214 detractors describing the same failed import step, up 40 percent this month." The rule to hold onto is traceability: any theme you are shown should link back to the exact responses that formed it, so you can verify it in seconds before taking it to a roadmap meeting. That is the job NPS software should be doing, and it is where most of the category still stops short.

The strongest version of this joins the comment to behavior. A detractor who wrote "too complicated" and never finished onboarding is a different problem from one who wrote "too complicated" after two years of daily use. The words only mean something next to what the person actually did.

How often should you ask?

The quarterly NPS blast is fading, and deservedly. It produces low response rates, a noisy score, a respondent pool skewed to the extremes, and a long lag between a problem appearing and you hearing about it.

Continuous sampling works better. A small slice of eligible customers sees the question each week, nobody is asked more than once or twice a year, and you get a rolling score you can genuinely trend plus a steady stream of comments to analyze. When a regression ships, it shows up in the feedback within days rather than at the end of the quarter.

Closing the loop is where the score actually improves

Measuring NPS changes nothing by itself. The improvement comes from the unglamorous operational work behind it: routing detractor responses to someone who follows up, fixing the themes that repeat, and telling the customers who complained that you fixed the thing they complained about. That last step is skipped almost universally, and it is the one that converts detractors into promoters.

That is a customer experience operations problem as much as a product one, and it is worth treating it as a real workflow with owners and SLAs rather than a dashboard someone glances at. Teams that get serious about closing the loop with the customers who complained tend to see the score follow, because they are fixing causes rather than watching an average.

The short version

Above 0 is good, above 20 is favorable, above 50 is excellent, above 80 is world class. Now ignore that and do the useful thing: segment your own score, sample it continuously instead of quarterly, read every comment, cluster the reasons, tie them to what those customers did in the product, fix the top theme, and tell the people who raised it. The number will take care of itself. If you want the wider picture of how NPS sits alongside the other satisfaction metrics, our comparison of CSAT, NPS and CES covers when to reach for each.