Skip to content

Asking Good Questions Is a Skill You Can Learn

The way you ask shapes the answer you get, from people and from tools. What research says about follow-up questions and how to ask with clear context.

A rounded question-mark shape of stacked orange blocks with a smaller indigo shape branching off it, orange and indigo on dark slate

You ask a teammate, "Any update on the report?" and get back, "Still working on it." You type "explain this" into a chatbot and get three paragraphs that explain everything except the part you were stuck on. In both cases the other side answered the question you asked. The trouble is that it was not quite the question you meant. The good news: asking well is not a talent some people are born with. It is a skill, and you can practise it.

Questioning is a skill most of us were never taught

In a 2018 Harvard Business Review article, The Surprising Power of Questions, Alison Wood Brooks and Leslie K. John point out a gap. Professionals such as litigators, journalists and doctors are taught how to ask questions as an essential part of their training. Yet few executives think of questioning as a skill that can be honed, or think about how their own answers could make a conversation more productive.

That gap is not only a problem for executives. Most of us learned to answer questions at school, in exams and assignments, and almost nobody sat us down to practise asking them. So we ask the way we always have, and we accept whatever answer comes back.

Brooks and John argue that this is a missed opportunity. According to the summary of the article on the Harvard Business School site, questioning spurs learning and the exchange of ideas, and it builds trust among team members. They name several techniques that make questions work better: favour follow-up questions, know when to keep questions open-ended, get the sequence right, use the right tone, and pay attention to group dynamics. The HBR summary adds a hopeful note: by asking questions we naturally improve our emotional intelligence, which in turn makes us better questioners.

The power of the follow-up

The first technique on that list has some of the clearest evidence behind it. A 2017 paper in the Journal of Personality and Social Psychology, It doesn't hurt to ask: Question-asking increases liking by Karen Huang, Michael Yeomans, Alison Wood Brooks, Julia Minson and Francesca Gino, looked at real conversations between pairs of people.

Across three studies of live conversations, the researchers found a consistent pattern: people who asked more questions, particularly follow-up questions, were better liked by their conversation partners. When people were told to ask more questions, their partners saw them as more responsive, which the authors describe as listening, understanding, validation and care. The team also studied face-to-face speed dating, and found that people who asked more follow-up questions during their dates were more likely to get a yes to a second date.

One more finding is worth keeping in mind: people did not expect asking questions to make them more liked. That may be why we ask fewer than we could. (In 2025 the journal published a correction after an independent audit of the paper. It fixed several minor reporting errors and confirmed that all of the paper's conclusions are valid.)

Why would a follow-up matter so much? A follow-up shows that you heard the last answer. "What made that part hard?" can only come from someone who listened to "That part was hard." It also tends to pull out the detail you actually need, the reason behind the status update or the step where the explanation lost you.

Clear context, for people and for tools

The same idea carries over to the tools you ask every day. Anthropic's prompting best practices for its Claude models put it plainly: think of Claude as a brilliant but new employee who lacks context on your norms and workflows, and the more precisely you explain what you want, the better the result. The same page says that explaining the reason behind a request gives the model more information about your goals and helps it give more targeted answers.

The page also offers a simple check that works far beyond AI. Show your request to a colleague who has little context on the task. If they would be confused, the model will be too. Anthropic writes this about its own models, but the test fits a message to a lecturer, a question in a team chat, or a ticket for the IT desk just as well. Anyone who was not in the room needs the room described to them.

Ways to practise this week

Pick one or two of these and notice how the answers change.

  1. Add one follow-up. In your next meeting or study group, after someone answers, ask one more question that builds on what they just said: "How did you decide that?" or "What would you do differently next time?"
  2. Choose open or closed on purpose. When you are exploring, ask open questions such as "What worries you about this plan?" When you need one fact, ask for exactly that fact: "Is the deadline Friday or Monday?"
  3. Run the colleague test before you send. Reread your question as if you had not been there. Does it say what you are working on, what you already tried, and what kind of answer you want?
  4. Lead with one line of context. Start with "I'm preparing a five-minute summary for my manager" or "I'm revising for a statistics quiz" before the question itself. The answer can then fit your situation instead of everyone's.

None of this takes longer than a few seconds. It just moves a little of the thinking to before you ask, which is usually cheaper than after.

Quick questions with Ask Anything

When the question is small and you want the answer now, Snipset has Ask Anything. You speak a quick question through its shortcut, and the local Ollama model answers briefly: Snipset asks it for one to three sentences, in the language you asked in. Because the answer is short, the habits above pay off: "What's the difference between mean and median, with one example?" gets you further than "statistics?". Web Search Grounding is off until you turn it on; when you do, DuckDuckGo is the default and TinyFish is an optional provider that needs your own key, as the TinyFish search post explains. Claude AI (cloud) is a separate choice with its own toggle for each feature, and Ask Anything only uses it if you switch that toggle on.

See Snipset plans and pricing →

Iwan Kurniawan

Written by

Iwan Kurniawan

@belajarcarabelajar

Keep Snipset in your Google results

Add snipset.belajarcarabelajar.com as a preferred source on Google. Snipset posts then show up more often when you search, with a preferred badge in AI Overviews.

Add Snipset on Google