Why questionnaires fail on sensitive topics

The format problem

It is not a problem of method. The assessments are valid, the scales reliable, the norms up to date. The problem lies deeper: in the format itself. A questionnaire invites simplification. Five levels from “does not apply at all” to “applies completely” – that is enough for a number, but not for the story behind it.

In non-critical contexts, this is not a problem. Anyone asked about job satisfaction ticks a box and moves on. But with topics that create uncertainty – safety culture, strain, leadership problems, compliance breaches – something else happens: people round off. Not because they are lying, but because the format does not allow for nuance. The three instead of the two. The neutral middle instead of the uncomfortable truth.

The result is data sets that look methodologically clean and say too little in substance. Teams that, according to the questionnaire, are “in the green zone”, while three floors down resignations are being prepared.

What a conversation does differently

When people talk instead of ticking boxes, two things happen at the same time. First, they become more precise. “Rather dissatisfied” becomes “the team leader changed the targets three times last quarter without involving us”. That is not a softer data point – it is a more precise one. And it can be mapped onto a scale if you want.

Second, they add to what they say. In conversation, people voluntarily share context that does not appear in any item. In the pilot on workplace health, participants on average provided significantly more relevant additional information than in the questionnaire – suggestions for improvement, pointers to structural problems, concrete descriptions of situations.

This context is the real added value. Not because the scale scores are missing, but because the additional information makes the difference between “we have a communication problem” and “we know where it comes from”.

The practical consequence

For HR departments and organisational developers, this means a different calculation. The classic route: roll out a questionnaire, interpret the results, plan a workshop, carry out a follow-up survey, derive measures. The new route: conduct an AI interview, receive both levels of results at once, go straight into planning measures.

This does not just save time. It changes the quality of decisions. Those who have concrete descriptions of situations instead of just scale averages design measures that take effect – not measures that look good.

When a questionnaire is enough

To avoid any misunderstanding: there are uses for which a classic questionnaire is exactly the right format. Regular short pulse surveys with three questions. Large samples where only distributions matter. Benchmarking across locations when the context is known.

The conversational format is worthwhile where context is missing and the consequences are serious. Safety culture. Leadership development. Strain assessments in phases of change. Insider threat assessment. Compliance climate. Wherever the gap between what people tick and what they actually experience makes the difference.

Further reading

If you are about to run a survey and are unsure whether a questionnaire or an AI interview is the right format – talk to us. We advise with an open mind and help you choose the format.