How to Learn to Ask Questions: The Definitive Guide to Interviewing Customers, Leading Teams, and Making Better Decisions

A practical guide to learning how to ask questions: why we ask badly, 7 types of questions with a table, the PADA method, templates for customers and 1:1s, common mistakes, and a 30-day plan.

Why Asking Is the Most Underestimated Skill

Asking well is cheap, and its return is asymmetric. A question takes ten seconds and can save you six months of development. It requires no budget, no permission, no tool. It requires method.

It's also the skill most often confused with another. Many people believe that asking questions is what someone who doesn't know does, and that's why they avoid asking in front of a client, an investor, or their own team. It's exactly the opposite. In professional settings, the person who asks with precision demonstrates that they understand the terrain well enough to know where the gap is. The naive question reveals ignorance; the precise question reveals mastery.

There's even evidence that asking improves how you're perceived. A study by Harvard researchers published in 2017 (It Doesn't Hurt to Ask: Question-Asking Increases Liking, Huang, Yeomans, Brooks, Minson, and Gino) found that people who asked more questions — and especially more follow-up questions — were rated as more likable by their conversational partner, both in lab conversations and in real speed dating. The intuition that asking too many questions is annoying turned out to be inverted.

And yet we still talk more than we ask. For three reasons worth facing head-on.

Why We Ask Badly: The Five Factory Defects

1. We ask to confirm, not to discover

This is the root failure, and it has a name: confirmation bias. When you already have an idea, your brain doesn't look for the truth, it looks for approval. And so you formulate questions that can only return a yes.

"Don't you think an app to manage subscriptions would be useful?" is a question designed to get a yes. "How many subscriptions do you have right now, and how do you keep track of them?" is a question designed to discover reality. The first one gives you encouragement; the second one gives you data.

The warning sign is simple: if you know in advance the answer you want to hear, you're not asking, you're seeking permission.

2. We ask about the future instead of the past

"Would you pay for this?" is the most useless question in entrepreneurship. Nobody knows what they'll do in the future, and everyone imagines themselves as a more disciplined, more rational, and more generous version of who they are. Answers about the future are well-intentioned fiction.

The past, on the other hand, has already happened and leaves a trace. "When was the last time you paid for something like this? How much? What made you decide that day?" are verifiable questions. Past behavior isn't a promise: it's evidence.

3. We contaminate the question with the answer

Leading questions embed the answer in the phrasing. The classic experiment by Elizabeth Loftus and John Palmer in 1974 demonstrated this uncomfortably: people who watched the same video of a car accident gave different speed estimates depending on the verb used in the question. Those who were asked how fast the cars were going when they "smashed" gave higher numbers than those who were asked when they "contacted". The question didn't measure memory: it modified it.

The same thing happens in business. "The current process is a mess, isn't it?" manufactures the mess. "Tell me how the current process works, step by step" describes it.

4. We ask to look good

Social desirability works in both directions. Your counterpart wants to look good to you and softens the truth. But you also want to look good to them, and that's why you don't ask the uncomfortable question: the one about price, about budget, about "what would make you not buy this?", about "are you thinking about leaving the team?".

The question you're afraid to ask is, almost always, the only one that matters.

5. We ask without leaving room for the answer

This is an execution failure, not a design one. You ask a good question, a two-second silence opens up, you get nervous and fill it: you rephrase, qualify, give examples, suggest options. You've just turned an open question into a multiple-choice test.

Education researcher Mary Budd Rowe documented in the 1970s that teachers waited on average about one second after asking a question before intervening, and that extending that wait to about three seconds produced noticeably longer and more elaborate responses. The finding translates directly to a meeting room or a customer interview. Silence isn't a void to fill: it's part of the question.

The Anatomy of a Good Question

A question that works meets four conditions. You can use this as a checklist before an important conversation.

It's open when it seeks territory and closed when it seeks a fact. There are no good open questions and bad closed questions: there are poorly chosen questions for their purpose. "How do you keep track of expenses?" opens. "How much did you pay last month?" closes. You need both, in that order.

It's singular. One question, one idea. "What tools do you use and how much do they cost and what would you like to change?" guarantees they'll only answer the last part or the easiest one.

It's neutral. It contains no adjectives that push ("is it frustrating...?", "isn't it slow...?"). Let the other person supply the adjective. If the problem is frustrating, they'll say so. And if they have to be the one to tell you, it's real.

It's concrete and anchored in time. "How do you manage your team?" invites a philosophy. "Tell me what you did last Monday when two people asked you for things at the same time" invites a fact.

And one contextual condition that encompasses them all: the question must be answerable without social cost. If answering it honestly makes the respondent look bad, you'll get a polite lie. Part of the work of asking is setting up the conditions beforehand so that the truth is cheap.

The Seven Types of Questions (and When to Use Each)

This is the toolkit. Most people only use two.

#	Type	What it's for	Example	Risk if you overuse it
1	Opening	Open territory without steering, see where the other person leans	"Tell me about a typical day at your work"	Scattered conversation, no actionable data
2	Concretizing	Turn a generality into a verifiable fact	"When was the last time that happened? What exactly did you do?"	Sounds like an interrogation without prior trust
3	Follow-up	Dig deeper into what they just said, a sign of real listening	"And what else?" / "What do you mean by 'complicated'?"	None relevant. It's the most underused one
4	Mirror	Repeat their last words in a questioning tone so they elaborate	Them: "...and in the end we dropped it." You: "You dropped it?"	Gets tiring if you do it every turn
5	Causal	Seek the mechanism behind the symptom	"What made that happen?" / "And why do you think it occurs?"	A direct "why?" can sound accusatory with people
6	Contrast	Force prioritization and reveal the real criterion	"If you had to choose between solving A or B, which one and why?"	Creates false dilemmas if the options aren't real
7	Calibrated	Shift the problem to the other party without confronting	"How are we supposed to do this with the current budget?"	Manipulative if used to corner, not to understand

Calibrated questions come from the work of Chris Voss, former FBI negotiator: open-ended questions that start with "how" or "what" and that invite the other person to solve your constraint out loud, instead of defending against an accusation. They work just as well negotiating a contract as they do asking your team for a realistic deadline.

A note on type 5. The classic 5 Whys — popularized in the Toyota production system by Taiichi Ohno — is excellent for processes and systems: you ask "why" five times in a row until you move from the symptom to the structural cause. But applied to people, a "why did you do that?" triggers defensiveness immediately. With people, replace "why?" with "what led you to...?" or "what were you seeing at that moment?". Same information, zero armor.

The PADA Method: Four Phases for Any Important Conversation

Asking well isn't about improvising brilliant questions. It's about having a structure and filling it.

P — Prepare

Before the conversation, write down two things: what decision you'll make with what you learn and which three of your beliefs you want to put at risk.

The first prevents the tourist interview, the one where you come away with a lot of pleasant conversation and no consequence. If there's no decision on the other side, the conversation doesn't need to happen yet.

The second is the key to everything. Write down your assumptions as falsifiable statements: "I believe freelancers waste more than two hours a month reconciling invoices", "I believe my best developer is bored, not burned out". Your goal in the conversation isn't to confirm them: it's to try to tear them down. If you come away with all three intact and not a single surprise, you probably didn't ask, you presented.

A — Open

The first minutes aren't for your important questions. They're for lowering the cost of telling the truth.

Three moves that work: frame the purpose honestly ("I'm not here to sell you anything, I'm trying to understand how this really works"), legitimize the uncomfortable answer in advance ("the most useful thing you can tell me is what doesn't work"), and start on the easiest terrain — the everyday and descriptive — before getting into sensitive territory.

Edgar Schein called this humble inquiry: asking from the position of someone who genuinely doesn't know and needs the other. It's not a courtesy technique, it's a condition for information to flow upward. Teams where the boss always asks from the position of someone who already knows the answer stop bringing them information — and what they stop bringing is usually exactly what they needed to hear.

D — Drill Down

This is where the real work happens, and it almost always consists of the same repeated move: from assertion to fact.

Every time someone gives you a generality, take it down a notch. It works like this:

— Yeah, invoice management is a pain. — What costs you the most about it? (concretizing) — Mostly reconciling what I got paid with what I invoiced. — When was the last time you did that? (temporal concretizing) — Last Friday. — Tell me about that Friday: what did you open first? (reconstruction) — Well, the bank, I exported a CSV, pasted it into a spreadsheet... — And how long did it take? (data) — About three hours. — And what else happened that day because of that? (follow-up) — I didn't call a client I had pending.

In six questions you've gone from an adjective ("pain") to a concrete opportunity cost (three hours and a lost sales call). The first answer would have led you to build a generic invoice manager. The sixth tells you that what you're selling is getting their Friday back.

The operating rule: never accept an adjective as data. Complicated, slow, expensive, frustrating, insufficient — all of them are doors, not conclusions. Behind each one is a fact with a date, a duration, and a consequence.

A — Anchor

Always close with three moves that almost no one makes.

Summarize and ask for correction. "Let me see if I understood: what hurts the most isn't invoicing, it's the month-end reconciliation, and it costs you about three hours. Is that right, or am I off?" This does two things: it verifies your understanding and gives them explicit permission to correct you, which is when the most valuable nuances surface.

Ask about what you didn't ask. "What did I ask about that wasn't important? And what should I have asked you but didn't?" It's the question that improves all your future conversations.

Ask for a concrete next step. If it's a customer: "who else should I talk to?" or a real commitment (a trial, a deposit, a date). If it's your team: "what do you expect from me after this conversation, and by when?". Without a next step, the conversation was entertainment.

Asking Customers: How Not to Fool Yourself

This is the terrain where asking badly is most expensive, because the error doesn't show up until months later.

The essential reference is The Mom Test by Rob Fitzpatrick. Its premise: you must formulate questions so good that even your mother can't lie to you with them, no matter how much she loves you. His three rules, summarized:

Talk about their life, not your idea. As soon as you mention your idea, the conversation stops being information and becomes courtesy.
Ask about concrete facts from the past, not opinions about the future.
Talk less and listen more. If you're talking more than 25% of the time, the interview is yours, not theirs.

Here's the operational translation:

❌ Don't ask	✅ Ask
"Do you think it's a good idea?"	"How do you solve this today?"
"Would you pay $30 a month for this?"	"How much do you spend today solving this, in money or in hours?"
"Would you use an app that did X?"	"What have you already tried for this? Why did you stop?"
"How often does this happen to you?"	"When was the last time? Tell me about that day"
"What would you think if we added Y?"	"If this didn't exist tomorrow, what would you do?"
"Is this important to you?"	"What other things did you stop doing because of this?"
"Would you like it faster?"	"What happened the last time it was delayed?"

And three signs your interview is getting away from you:

They're flattering you. "What a great idea", "I love this", "this will definitely work". Compliments are noise. When they appear, redirect: "thank you, but tell me how you do it now".

It's all hypotheticals. If the answers come in the conditional ("I would use", "it would be nice if"), you're not collecting data, you're collecting imagination.

Nobody has said anything that surprised you. It's the most reliable sign that you asked leading questions. An interview without surprises is an interview where only you talked, even if the other person used more words.

One last criterion that Fitzpatrick calls commitment and advancement: the information that matters comes accompanied by something that costs the other person. Time (a second meeting scheduled), reputation (a presentation to their boss), or money (a deposit, a letter of intent). A free "I love it" is worth less than a "I'm not interested" with a reason.

Asking Your Team: 1:1s That Are Good for Something

The average 1:1 is a status report disguised as a conversation. It starts with "how are you doing?", continues with a review of tickets, and ends with "let me know if you need anything". Nobody says anything. Everyone goes back to work.

The problem is that "how are you doing?" is a question with no surface. There's nothing to grab onto, so the default answer is "fine". And "fine" is what people say right before they quit.

Michael Bungay Stanier, in The Coaching Habit, proposes a short, highly reusable repertoire. Adapted:

Moment	Question	What it unlocks
Kickoff	"What's on your mind?"	Lets them choose the topic; what's real usually comes out, not what's on the agenda
Go deeper	"And what else?"	The best effort-to-value ratio of all. The third answer is usually the good one
Focus	"What's the real challenge here for you?"	Separates the problem from that person's problem
Desire	"What do you want?"	Forces them to name the request they were circling around
Help	"How can I help?"	Prevents you from solving the wrong problem on your own
Strategy	"If you say yes to this, what are you saying no to?"	Makes the opportunity cost visible
Close	"What was most useful to you about this conversation?"	Consolidates learning and tells you what to repeat

On "and what else?" it's worth insisting. The first answer someone gives is the rehearsed one. The second is the real one. The third is the one even they didn't know they had. Most conversations cut off at the first because the boss is already thinking about their solution.

And that's the other typical failure: asking as an excuse to answer. If you ask the question but already have the answer prepared, what you're doing is a test, not a conversation. People notice immediately and adjust what they say to what they think you want to hear. From there on, your 1:1s only confirm what you already thought — and your team becomes an expensive mirror.

Three more questions, for specific moments:

When someone seems disengaged: "What part of your work would you do even if you weren't paid, and what part would you only do if they paid you double?"
When you want real feedback about yourself: "What is it that I do that makes your job harder?" (much better than "how am I doing?", which only has one socially acceptable answer).
When someone brings you a problem: "What would you do if I weren't here?" They almost always already have the answer; what they're looking for is authorization.

Asking in Meetings and Decisions: Surfacing the Assumptions

In a team decision, the enemy isn't disagreement: it's premature agreement. Everyone nods, no one has said out loud what they're basing it on, and three weeks later it turns out everyone understood something different.

Chris Argyris described this with the Ladder of Inference: we start from observable data, select some of it, assign it meaning, draw conclusions, adopt beliefs, and act — and in meetings we usually start arguing from the top rung, defending conclusions without ever having compared the starting data. Two people can argue for an hour when in reality they weren't looking at the same facts.

The questions that bring people down the ladder:

"What makes you think that?" (asks for the data underneath the conclusion)
"What would have to be true for this to work?" (turns an opinion into a list of verifiable assumptions — Roger Martin's favorite question in strategic decisions)
"What evidence would make us change our minds?" (defines the failure criterion before starting)
"Who disagrees and hasn't said so?" (makes silent dissent explicit)
"If this goes wrong in six months, what will be the most likely reason?" (Gary Klein's premortem, the most efficient way to surface risks without anyone looking like the party pooper)

And for incidents and errors, a substitution that changes an entire team's culture:

❌ Blame question	✅ System question
"Who broke it?"	"What made this possible?"
"Why didn't you test it?"	"What would you have needed to catch it earlier?"
"How did you miss that?"	"What seemed like a reasonable thing to do at that moment with the information available?"

The first column gives you a culprit and no improvement. The second gives you an improvement and no culprit. Only one of the two prevents the incident from recurring.

The Most Common Mistakes, with Their Corrected Versions

Mistake	Example	Fix
Double question	"What do you use and how much do you pay?"	Split them. Always
Leading question	"Isn't it frustrating to have to...?"	"What's that process like for you?"
Hypothetical	"Would you use this?"	"What do you use today?"
With jargon	"How do you optimize your activation funnel?"	"What do you do when a new user comes in?"
Prematurely closed	"Would a weekly report work for you?"	"What do you need to know, and how often?"
Interrogation	Ten questions in a row with no reaction	Intersperse paraphrases: "so what you're saying is..."
Filling the silence	Question + three example answers	Ask. Count to three. Stay quiet
Asking to show off	"Have you considered the impact on CAC?"	If you already know the answer, it's not a question
Not asking the uncomfortable thing	Avoiding the topic of price	Name the discomfort: "this is the awkward question, but..."

Listening: The Half Nobody Talks About

A good question with bad listening is a wasted question. Three things you can control:

Take literal notes, not interpreted ones. Write down what they said in their words, in quotes. As soon as you summarize while writing, you've already filtered through your assumptions, and the customer's exact words are pure gold for your copy, your pitch, and your product. "Closing the month" is an expression already worth more than any headline you can come up with.

Listen to what they don't say. The topics they circle without entering, the answers that suddenly shorten, the change in tone when a certain person is mentioned. That's usually where the expensive stuff is. You can point to it without invading: "I noticed you changed the subject quickly there, is there something more?".

Separate listening from deciding. While someone is talking, your brain wants to evaluate, refute, and plan the response. Postpone that. Practical rule: don't evaluate anything until you've closed the conversation and reread your notes the next day. Decisions made inside the conversation are almost always the ones you had before starting it.

A 30-Day Plan to Train This

Learning to ask better doesn't happen by reading lists of questions. It happens by doing one different thing each week until it stops being deliberate.

Week 1 — The silence. Single goal: after every question you ask, count to three before saying anything. Nothing else. It's uncomfortable for four days and then becomes invisible. Notice how much what they tell you changes.

Week 2 — "And what else?". In every work conversation, use a follow-up question at least twice before changing the subject. Forbidden to move on to your next prepared question without having dug deeper into the previous answer.

Week 3 — From adjective to fact. Every time someone uses a vague word (hard, slow, expensive, bad), ask for the fact: when, how much, what exactly happened. Count how many times you manage it.

Week 4 — The uncomfortable question. One per day. The one you've been avoiding: price, the real deadline, feedback about yourself, the real reason for a no. Formulate it by naming the discomfort ("I'm going to ask you a direct question...") and notice that almost never does anything bad happen.

At the end of the month, do an interview or a 1:1 and evaluate yourself with three numbers: the percentage of the time you talked (goal: less than 30%), the number of concrete facts with dates you came away with (goal: more than five), and the number of times you surprised yourself (goal: at least one). If the third number is zero, go back to week 1.

Frequently Asked Questions

How many questions should I bring prepared to a customer interview? Between five and eight at most, and none of them is a script. They serve as a safety net, not an itinerary. If the conversation goes down an unexpected and fertile path, abandon your list: the unexpected is exactly what you went looking for.

How many interviews do I need to validate something? There's no magic number, but a useful criterion is saturation: stop interviewing when three or four consecutive conversations don't give you any new category of problem. In practice this usually happens between the tenth and twentieth interview within the same segment. Watch out for that last part: saturation is per segment, not total. Changing profiles resets the counter.

What if my customer doesn't know what they want? That's the normal situation, not the exception. That's precisely why you don't ask what they want: you ask what they do. People are a terrible source of solutions and an excellent source of problems and past behavior.

Doesn't asking a lot make me look insecure in front of my team or a customer? It depends on the type of question. Asking "what should I do?" projects insecurity. Asking "what would have to be true for this to work?" projects judgment. The difference isn't how much you ask, but whether your questions have direction.

Does this work in surveys or forms? The same principles apply — avoid leading, double, and jargon-filled questions — but you lose the most valuable part: the follow-up. A survey can't say "and what else?". Use them to measure something you already understand, never to discover something you don't yet understand.

How do I ask about sensitive topics like price or salary? By naming the discomfort before asking and normalizing the difficult answer: "I'm going to ask you a direct question and it's fine if you'd rather not answer". And ask for facts, not disposition: "how much did you pay?" is easier to answer honestly than "how much would you pay?".

In Summary

Asking well isn't a talent of sociable people. It's a method with recognizable parts: prepare the decision you'll make and the assumptions you want to tear down, open by lowering the social cost of the truth, drill down through every adjective until you reach a fact with a date, and anchor with a correctable summary and a next step.

If you take away only one idea, let it be this: the goal of a good question isn't to get an answer, it's to change what you believe. A conversation in which you didn't change your mind about anything wasn't a conversation. It was a presentation with turns.

And the uncomfortable part of all this is that the skill gets trained in the questions you don't want to ask. The one about price. The one about feedback about yourself. The one about "why didn't you buy from me?". Start with one this week.

If you're interested in the decision-making side, this article reads well alongside Cognitive Biases for Entrepreneurs and Freelancers: good questions are, to a large extent, an antidote to the biases described there.