"We're a 42, the published industry average is 35" is a satisfying sentence to put in a slide, and it's worth being suspicious of exactly how satisfying it feels. A benchmark comparison is only as good as how comparable the two numbers actually are underneath the headline figure, and cross-company benchmarks - the kind published in industry reports, vendor studies, and "average NPS by sector" roundups - are frequently measuring something close to, but not quite, what you measured yourself. That gap doesn't make benchmarks useless. It makes them worth using carefully, for the specific things they're actually good for, rather than as a scoreboard.
Table of Contents¶
- Why Two Numbers Aren't Always Comparable
- What Actually Needs to Match
- Average Benchmarks vs. Percentile Benchmarks
- The Benchmark That Matters More Than Any External One
- Benchmarking Within Your Own Organization
- When an External Benchmark Is Genuinely Useful
- A Worked Example
- FAQ
Why Two Numbers Aren't Always Comparable¶
The same metric name can hide meaningfully different measurements. Two companies both reporting "NPS" might be asking the question at different points in the customer relationship, to different slices of their customer base (everyone versus only recently active users), with different response scales presented slightly differently, and pulling from response rates that differ enough to change who's actually represented in the number. None of this is visible in a single published average, and all of it can move a score by several points on its own, independent of anything about how customers actually feel. Our guides on survey metrics by industry and NPS analysis both touch on this same caution - a benchmark tells you roughly where things tend to land in a category, not a precise target your specific number should be judged against.
What Actually Needs to Match¶
A genuinely comparable benchmark needs to match on more than just the metric's name. Question wording matters - even small differences in phrasing or scale labeling can shift a score by a meaningful margin. Timing in the customer or employee relationship matters - a metric collected right after onboarding reads differently than the same metric collected a year in. Audience definition matters - "all customers" and "customers active in the last 90 days" are not the same population, and the second one is almost always going to score better, since it's implicitly excluded some of the people most likely to be dissatisfied. And industry or company-size context matters more than most published benchmarks account for - a benchmark blending enterprise and small-business respondents together isn't a fair comparison point for a company that's purely one or the other. When a published benchmark doesn't specify these details clearly, that's a signal to treat it as a loose directional reference rather than a precise bar to clear.
Average Benchmarks vs. Percentile Benchmarks¶
Most published benchmarks report a single average, and a single average hides how spread out the underlying companies actually were - "the average NPS in this category is 35" could describe a tight cluster where almost everyone scores between 30 and 40, or a wide spread where a handful of standout companies in the 60s are dragging the average up while most of the category sits in the teens and twenties. Those are very different competitive landscapes to be benchmarked against, and a single average number can't tell you which one you're actually in.
A percentile-based benchmark - "the top quartile in this category scores above 48, the median is 32, the bottom quartile is below 18" - is more informative when it's available, because it tells you where you'd need to land to be genuinely ahead of the pack, not just above an average that a few outliers might be inflating or deflating. If you only have access to a single average figure, it's worth treating your own position relative to it more loosely than a percentile-based comparison would justify - "roughly in the neighborhood of typical" rather than a precise ranking claim.
The Benchmark That Matters More Than Any External One¶
Your own number from your own last measurement period, collected the same way, is a more trustworthy benchmark than almost any external comparison you'll find, precisely because it's guaranteed to be measuring the same thing the same way. This is the logic behind tracking a metric over time - see our guide on tracking a metric across waves for the mechanics - and it's worth treating as the primary benchmark, with external comparisons as a secondary, lower-confidence reference point rather than the main event. A team that improves steadily against its own history is making real progress regardless of where an external benchmark happens to place them; a team fixated on beating an external number that isn't measuring quite the same thing can end up either falsely reassured or falsely alarmed by a comparison that was never quite fair to begin with.
Benchmarking Within Your Own Organization¶
Between your own trend over time and an external industry number sits a third, often-overlooked option: comparing results across departments, regions, or teams within your own organization, using the exact same survey. This carries most of the fairness advantage of your own historical trend - same question wording, same timing, same methodology - while still giving you a genuine comparison point rather than just tracking against your own past self. A regional sales team's customer satisfaction score sitting meaningfully below every other region's, on the same survey fielded the same way, is a far more trustworthy signal than the same gap measured against an external published benchmark of unknown comparability.
This kind of internal benchmarking works best when the units being compared are genuinely similar in what they're being asked to deliver - comparing an enterprise sales team's numbers against a self-serve team's on the same survey can reintroduce the same apples-to-oranges problem external benchmarks have, just at a smaller scale. Grouping comparisons by genuinely similar internal units, the way you would when cross-tabulating survey data by any other meaningful segment, keeps this internal benchmark honest.
When an External Benchmark Is Genuinely Useful¶
External benchmarks earn their keep for orientation, not precision - a rough sense of "is this number in a plausible range for our category" when you have no internal history yet to compare against, particularly useful the first time you run a survey and have nothing of your own to benchmark against. They're also useful for spotting a genuinely extreme outlier - a score dramatically below any published range for your category is worth investigating regardless of exactly how comparable the benchmark's methodology is, since a gap that large is unlikely to be fully explained away by measurement differences alone. What they're not especially useful for is fine-grained comparison - treating a 3-point gap from a published average as meaningful is asking more precision from a benchmark than most published benchmarks can actually support.
A Worked Example¶
A mid-sized B2B software company sees a published industry report listing average NPS at 38, against their own measured score of 29, and initially reads this as a real, concerning gap. Digging into the report's methodology page reveals it surveyed enterprise companies exclusively, with a median company size roughly ten times larger than theirs, and measured NPS specifically among customers active in the last 30 days - a narrower, more engaged slice than the company's own all-customer survey population. Rather than chasing the published number directly, the team instead pulls their own NPS trend over the last four quarters, which shows a steady climb from 24 to 29 - real, measurable progress against a fair, apples-to-apples comparison. They keep the external benchmark as rough context in the appendix of their report, but lead the actual narrative with their own trend, which is the number that was actually telling them something true.
FAQ¶
Are industry benchmark reports worth reading at all?
Yes, for general orientation and spotting genuine outliers - just don't treat a small gap from a published average as meaningful without understanding how that average was actually measured, since minor methodology differences can easily account for a few points either way.
Is it possible to benchmark fairly against a competitor?
Only if you can verify their methodology matches yours closely enough on question wording, timing, and audience definition - which is rare with published competitor numbers, since methodology details are often left out of marketing-facing benchmark claims entirely.
Should I stop comparing to industry benchmarks altogether?
No - just rebalance how much weight the comparison carries. Your own historical trend deserves to be the primary story in any report; an external benchmark is a reasonable supporting reference, not the headline.
How do I make my own benchmark more reliable over time?
Keep the measurement itself consistent wave over wave - same question wording, same timing in the relationship, same audience definition - which is exactly what makes your own trend a fairer, more trustworthy comparison than most external ones in the first place.
Is a percentile benchmark always better than an average benchmark?
When it's available, yes - it tells you where you'd need to land to be genuinely ahead, rather than just above or below a single average that outliers can distort. Not every published benchmark reports percentiles, though, so an average is often what you actually have to work with.
Can I benchmark different departments or teams against each other internally?
Yes, and it's often a more trustworthy comparison than an external one, since the survey methodology is guaranteed to match. Just make sure the units being compared are doing genuinely comparable work - comparing very different team types on the same survey can reintroduce the same fairness problem external benchmarks have.
For more on building a trustworthy trend and choosing the right comparison metric, see Tracking a Metric Over Time and Beyond Averages: The Professional's Guide to Survey Analysis.