BenchmarksNPS

What's a good NPS for an internal tool?

Short answer: anything above 0 is solid, above +20 is genuinely good, and comparing your expense tool to Apple's +60 is how internal teams end up demoralized by a number that was never calibrated for them.

July 21, 2026 · 5 min read

Why internal NPS runs structurally lower

NPS asks "how likely are you to recommend this?" — a question invented for products people chose. Internal tools start with three structural handicaps:

No self-selection. A consumer brand's respondents already voted with their wallet; unhappy customers left before the survey. Your users include everyone the tool frustrates, because they can't leave — visibly, anyway (they leave invisibly).
The recommendation is hypothetical. "Would you recommend the procurement portal?" To whom? Enthusiasm has nowhere to go, so 9s and 10s are rarer by construction.
The tool inherits the process's reputation. Users score the expense policy, the approval chain, and the tool as one experience. Some of your detractors are angry at a rule, not a screen.

Realistic ranges

Internal-tool NPS	Read it as
Below −20	Active hostility — expect visible workarounds and shadow IT. Treat as an incident.
−20 to 0	The unremarkable middle where most internal software lives. Not a crisis; not a résumé line.
0 to +20	Good. Users' needs are met and some are enthusiastic. Above the bar for most estates.
+20 to +40	Genuinely strong — worth studying what this team did and copying it.
Above +40	Exceptional for a tool nobody chose. Verify the sample before celebrating (see below).

Treat these as reading aids, not gospel — the honest truth is that public benchmarks for internal tools barely exist, which is exactly why calibrating against your own portfolio's distribution beats chasing a magazine number. Your best decision-instrument is relative: which tools sit at the top and bottom of your estate, and which direction is each moving?

The annual survey is the wrong instrument anyway

Most enterprises measure internal satisfaction once a year, in a 40-question omnibus, answered from memory weeks after the frustration happened. That yields a response rate skewed to the aggrieved and the dutiful, a score you can't attribute to any specific release, and a 12-month feedback loop for tools that ship monthly. The alternative that works:

In-product, in-the-moment: a 0–10 microsurvey shown inside the tool, while the experience is fresh.
Random sampling with cooldowns: a small slice of visitors (Morvero samples 0–25%, with a 7-day per-visitor cooldown) so the data flows continuously and nobody is nagged.
Anonymous: employees grade honestly when the grade can't be traced. (Why this matters legally too: anonymous by design.)
Trend over level: a weekly NPS trend line per product turns "our score is 12" into "our score dropped 9 points the week we shipped the new approval flow" — a sentence someone can act on.

Sample-size hygiene

NPS is a difference of proportions, and at internal-tool sample sizes it's noisy: at n=50, the score's margin of error is roughly ±20 points. Rules of thumb: don't quote a score below n≈30; don't compare two tools whose gap is under 15 points; and never celebrate or panic over a single month's movement on a small tool. Pair NPS with continuous star ratings — the volumes are 10–50× higher, which makes the trend line trustworthy at portfolio scale.

Where Morvero fits: NPS microsurveys with sampling and cooldowns, weekly trend lines per product, and star ratings riding the same widget — all anonymous, all on one portfolio dashboard. Start free and get your own baseline instead of borrowing B2C's.