Open-ended survey questions that get real reasons.

Open-ended survey questions ask people to answer in their own words. You give them a topic, such as their last order, rather than a fixed list of choices. Their reply tells you what stood out and gives you language your team would never have written.

The useful part is the experience behind that reply. “Unclear” names a problem, but leaves your team guessing about the fix. Underheard asks what felt unclear while the customer is still there. You hear the missing detail and carry it through to the report.

You can use an existing survey to choose a topic for deeper interviews. Invite the customers whose experience matters and ask them to describe a recent event.

A broad question leaves customers deciding what to explain and how much to type. On a phone, a single word is an easy way to finish. A question about a feature they never used asks them to guess. Repeated comment boxes give them the same task again.

Start by narrowing the question to something the customer remembers. Ask what happened on their last visit, order or support request. Underheard hears the answer and follows the detail they mention, so “slow” becomes the wait that interrupted their day.

Check completed responses as well as completion rates. You need an explanation you can act on. A submitted form with a one-word answer still leaves that work ahead.

Group answers by meaning, count each theme and read the comments behind it. For each finding, ask whether the customer described what happened or simply gave it a label. Keep different causes separate, even when people use the same word.

• Read in contextRead the question with the whole answer. “Expensive” can describe the total, a fee or the value someone expected.
• Group the reasonsGive each theme a clear meaning. Put billing confusion and delivery costs in separate groups when customers describe different problems.
• Count peopleSay how many people mentioned each theme and how many answered. One person can describe more than one issue.
• Check the exceptionsRead answers that challenge your main finding. They help you see which customers need a different change.

Underheard does this work on the spoken interviews. You review themes, counts and supporting quotes, then open the source conversations to check the conclusion.

Old comments help you choose what to investigate next. Read a sample, mark the missing details and decide whose experience would explain them. A theme called “confusing” is a starting point for a new question.

If your comments already name the broken step, use that evidence. When a comment stops at a label, invite customers to explain that part of the experience in an Underheard interview. Ask about a recent example and follow the words they use.

Keep the two sources clear in your readout. The survey tells you what people typed. The new interviews explain the reasons they describe when you ask further.

Ask about an event, then probe the detail the customer brings up. A useful probe asks for an explanation without suggesting your own cause. Use these examples to sharpen the questions in your guide.

• After a purchaseStart with “Tell me about placing your last order.” If they mention a surprise, ask “What did you expect instead?”
• After supportAsk “What happened when you last asked for help?” Follow a mention of waiting with “What did that delay affect?”
• After a product taskAsk “Walk me through the last time you tried to share a file.” Then ask about the step they found difficult.
• After a cancellationAsk “What was happening when you decided to cancel?” Follow the reason they give with a request for a specific example.

Keep each question focused on one thing. Let the customer finish before moving on. Underheard listens to the full reply and chooses a relevant follow-up from what was said.

Choose the decision first, then the people whose recent experience can explain it. Underheard drafts your guide from the brief. You edit it and try a conversation before opening the study. Give a short answer yourself and hear which detail the interviewer asks about.

Invite your own customers by link, or use Underheard’s panel with your quotas. Participants join on a phone or laptop, with nothing to install. Everyone is told at the start they are talking to an AI. Voice-to-voice. Text and video available.

A study takes a couple of hours to under a week. Underheard runs hundreds of customer interviews in a week. You spend your time choosing what to learn and deciding what the evidence means.

Match each proposed change to the reason customers described. A confusing billing period calls for clearer price wording. A surprise delivery charge calls for an earlier explanation of the total. Those are different decisions, even when both started with “unclear”.

Share an Underheard report link or export PowerPoint, PDF or Markdown. Keep the quotes and source questions beside the recommendation so colleagues can inspect the evidence. After changing the experience, ask customers about a new recent event and check what changed.

For more on the conversation itself, read about AI-moderated interviews. To hear individual customer reasons without assembling a group, explore focus group alternatives.