A score tells you what. A follow-up question tells you why.

Most feedback starts with a number. Rate your stay from 1 to 10. How likely are you to recommend us? How satisfied are you with your workplace? Numbers are easy to collect, easy to compare and easy to put in a chart.

They’re also hard to act on. Say your average drops from 8.1 to 6.4 on Friday nights. Something is clearly wrong. But what? The kitchen? The music? The staff? The team starts guessing, and every guess becomes a meeting.

The empty comment box

Many surveys try to solve this with an open field at the end: “Anything else you’d like to share?” It rarely works. By the time people get there, they’re done. The field stays empty, or gets a single word: “fine”, “ok”, “could be better”.

The problem isn’t that people have nothing to say. It’s that a general question at the end of a form doesn’t invite a specific answer.

What a follow-up does differently

A follow-up question reacts to what someone just said. It comes right after the score, and it’s about that score.

Three short exchanges, and the 6 has turned into something you can fix. Not “service is a problem”, but “the check takes too long when it’s busy”.

This works for a few reasons. The question is specific, so the answer is too. It shows that someone is paying attention, which makes people more willing to explain. And it’s asked while the memory is fresh, before the details fade into a general feeling.

Five rules for a good follow-up question

1. Build on their words. If someone mentions the wait, ask about the wait. Don’t jump to a question you prepared in advance.
2. Ask one thing at a time. “What went wrong, and how could we improve?” is two questions. People answer the easier one.
3. Stay neutral. “Was it the waiting time?” plants an answer. “What went wrong?” lets them tell you.
4. Ask about what happened, not about opinions in general. “What happened when you called us?” gets a story. “What do you think of our customer service?” gets a verdict.
5. Know when to stop. One or two follow-ups is usually enough. After that, say thank you. A conversation that drags on starts to feel like an interrogation.

Don’t skip the good scores

It’s tempting to only follow up on low scores. But a 9 deserves a question too. “What made it a good evening?” tells you what to protect. When you change things later, you’ll know what not to break.

Where AI helps

Asking a good follow-up question is easy when one person talks to one guest. It’s hard when hundreds of people answer every week, each with a different reason. That’s where AI is genuinely useful: it can react to each answer on its own terms, in the same tone, without a script that tries to predict every possible response.

It helps on the other side too. Two hundred short stories are hard to read in one sitting. Grouping them into themes, counting how often each one comes up and pointing to the actual answers behind each theme turns them into something you can act on.

What matters is that you stay in control. You decide what the survey is about and which topics the follow-up questions may touch. The AI asks within those limits, not beyond them.

Try it on what you already have

You don’t need to start over. Take the rating question you already ask and add one follow-up: “What’s the main reason for that score?” Run it for a month and compare what you learn with the month before.

If you’d like to see how it feels as a respondent, try the demo on our homepage. It uses fixed follow-up questions, but the idea is the same.