Fintra Feature

eNPS and pulse surveys, with anonymity enforced in the math

A hard k-anonymity floor means any driver, team slice, or manager cut answered by fewer than three people is mathematically suppressed: the aggregate for a group under the floor is never even computed. Add deterministic driver correlation, and you get a survey you can act on and an employee can actually trust. Anonymity isn’t a promise under the form; it’s a property of the computation.
Book a demoLog in

Pulse & eNPS

Example figures
eNPS
+34
promoters − detractors
RESPONSE
78%
640 / 820
k-FLOOR
k = 3
enforced in math
Manager relationship
4.2 / 5 · n=88
Compensation fairness
3.1 / 5 · n=64
Design team · career growth
n=2: suppressed
Top driver of eNPS (corr.)
recognition · r=0.61
Cut below k=3?
result withheld
Illustrative product view

K-anonymity that lives in the computation

Any breakdown that would expose an answer from fewer than three people (a single driver, a team slice, a manager’s direct reports) is suppressed before it is ever shown, because the aggregate for a group under the floor is never computed in the first place. It is not a display-time filter that a report export could bypass; it is enforced where the numbers are made.

Deterministic driver correlation

Each pulse computes which drivers move eNPS the most using a deterministic correlation: the same responses always produce the same ranking. So when the tool says “recognition is the top lever this quarter,” you can act on it without wondering whether a stochastic model would have reshuffled the order on a re-run.

Why “we promise it’s anonymous” isn’t the same

Culture Amp and 15Five protect anonymity by policy and by minimum-response thresholds applied when a report is rendered. That’s a promise about behavior. Fintra enforces the floor inside the aggregation itself: a stronger guarantee, because the small-group number simply does not exist to be leaked. For a survey, where candor depends entirely on trust, that difference is the whole point.

Questions people ask

What is k-anonymity in a survey?

It’s a floor on how small a reported group can be. With k=3, any driver, team, or manager cut answered by fewer than three people is suppressed, the aggregate for that group is never computed, so an individual’s answer can’t be isolated.

What happens to a team too small to report?

Its cell shows “suppressed (n<3)” and its responses roll up only into larger aggregates like the org-wide score. No one can drill into a below-floor group, so a two-person team’s answers stay unattributable.

How is this different from Culture Amp or 15Five?

Those tools protect anonymity by policy and apply minimum-response thresholds at display time. Fintra enforces the k-anonymity floor inside the computation itself, so the small-group number never exists to be exposed in the first place.

Is the driver analysis a black box?

No: driver correlation is deterministic. The same responses always yield the same ranking of what moves eNPS, so you can act on the top driver without worrying that a re-run would reshuffle it.

Can I keep years of survey history?

Honestly, not yet as a durable system-of-record: the survey engine currently runs in-process on a non-durable store. It’s a working product for running and reading pulses today; long-horizon retention is a roadmap item, and we’d rather say so.

Keep reading

One finance briefing a week

New guides, checklists and benchmarks, by email. Unsubscribe from any issue.

 

One brain for people, work, and money

See humans, AI agents, support, and finance decisions in a single governed workspace.

Book a demoLog in