Which tool to choose
The tool is the conversational assistant, to which you give the survey answers to bring out the themes. Before you paste them, though, assess the sensitivity: if the survey was anonymous, remove anything that could identify people, and bear in mind you're handing it to an external service. For sensitive topics, anonymize carefully. Ask for the themes, plus examples and the general mood, but never ask the AI to guess who wrote what. The tool serves to understand what the team thinks, not to unmask who said it.
How to do it
- Prepare the answers by removing names and details that identify people, especially if the survey was anonymous.
- Paste them and ask for the recurring themes, the general mood and the biggest points of tension.
- Ask it to turn the problems that emerge into concrete, realistic actions.
- Share the results with the team and what you intend to do: feedback that's collected and ignored destroys trust.
A concrete example
Luca, who managed a small team, had run an anonymous climate survey. Before analyzing it he removed every reference that could reveal who had written what. Then he asked the AI for the main themes: it turned out people felt poorly informed about decisions. The assistant helped him turn this into an action: a short weekly update to the team. Luca shared the results and the action he'd chosen, without ever trying to work out who had written what. The team saw that the survey actually served a purpose, and the next time they answered with more trust.
When it does NOT work (and how to fix it)
If you risk identifying people
In a small team, one detail is enough to reveal who spoke. Anonymize carefully before analyzing, and never share quotes that could be traced back to someone. Protecting people comes before the analysis.
If the analysis stays generic
"The team wants more communication" isn't enough to act on. Ask the AI to translate each theme into a concrete, feasible action. A survey is only useful if it leads to something that actually changes.
If you then do nothing with the results
This is the worst mistake: asking for feedback and ignoring it. The team notices, and the next time they won't put themselves out there. Use AI to reach realistic actions, and communicate what you'll do. Even a small change is worth more than a perfect analysis left in a drawer.
A tip from someone who really uses it
Remember that behind the answers there are people who trusted you, and treat that data accordingly. The temptation, with a survey in hand, is to use it to figure out "who's complaining" or to monitor: it's the fastest way to kill trust and ensure no one is ever honest again. AI makes analyzing extremely easy, but ease doesn't change the responsibility: anonymity respected, no hunt for who said what, and above all results that lead to real actions. A team gives honest feedback only as long as it sees that it serves to improve and not to surveil. Use the assistant to listen better, never to monitor more.
Frequently asked questions
Can I give the AI an anonymous survey?
Only after removing everything that identifies people, and with the awareness that you're handing it to an external service. The anonymity promised to the team must be protected: anonymize carefully before analyzing.
How do I turn the results into something useful?
By asking the AI to translate each theme into a concrete, realistic action. A survey is worthwhile if it leads to changes; the summary is only the first step toward decisions.
Can the AI tell me who wrote a comment?
No, and you mustn't ask it to. Hunting for who said what betrays the team's trust and drains the meaning out of any future survey. The analysis serves to understand the themes, not the people.
If I analyze the survey, have I listened to the team?
Only halfway. Analyzing is quick and easy with AI, but really listening means two more things: protecting whoever spoke up and doing something with what emerged. A survey analyzed brilliantly and then left in a drawer, or used to monitor, achieves the opposite of listening: it teaches the team that speaking up is pointless or risky. The AI's summary is a tool; real listening is the trust you build by respecting anonymity and turning feedback into visible actions. Stopping at the analysis is the most common way to gather feedback that, in the long run, no one will give you anymore.