Why your climate survey (and your eNPS) lies — and how real anonymity fixes it

Most engagement and climate surveys have the same quiet flaw: they ask people to be honest in a system where honesty can be traced back to them. So people do the rational thing — they answer safe. Detractors become passives, real problems become 'room to improve', and the dashboard turns a comforting shade of green.

You measure the fear, not the culture

MIT Sloan found toxic culture is 10.4× more predictive of attrition than pay — yet those same toxic environments are exactly where people least dare to say so on a form with their name attached (or attachable). The result is a cruel inversion: the worse the culture, the better the survey can look, because fear rises faster than the courage to report it.

Promising anonymity isn't enough

'Your answers are anonymous' is a policy — and policies can change, leak, or be quietly overridden by whoever runs the tool. People know this, so a promise doesn't fully unlock honesty. What unlocks it is when identifying an individual answer is not against the rules but impossible.

Anonymity you can't break by design

That's structural anonymity: the answer only ever increments an aggregate counter, and identity is used solely to avoid asking twice — the row that links a person to their answer is never written anywhere. Add a floor (nothing shown below 5 responses) and the honest answer to 'who are the detractors?' becomes: it can't be known, by anyone. That is the point at which people finally answer the truth.

It's the same reason a private ballot works. When Sabia asks eNPS or a climate question in Slack, the number reaches the company; who said what reaches no one. Measuring stops feeling like surveillance — and the data finally becomes worth acting on.

See structural anonymity in action →