Most conversation starters fail for the same boring reason: they ask for an inventory instead of an opinion. “What do you do?” gets you a job title. “Where are you from?” gets you a place name. Neither leaves any path back into the actual person. The good prompts — the ones that turn small talk into real talk in two minutes — share a structure. Once you see it, you can write your own forever.
Why most starters fail
A bad question makes the answerer feel like a database. Name, role, hometown, favorite food. They've answered all of those a thousand times, and the thousandth answer is the same as the first. There's no opportunity for them to say something they've been thinking about. There's no path to surprise.
Good questions are different. They invite someone to take a position, defend a taste, admit a contradiction, or describe a texture they don't usually put into words. Four shapes do most of the work.
The four shapes
1. Reversal questions
“What's something most people get wrong about [X]?” This works because almost everyone has a private grievance about how their world is misunderstood by outsiders. Doctors have one. Teachers have one. People with kids have one. Twenty-somethings have one. They've been waiting for someone to ask, and you just did.
The reversal version of the same question (“What's something everyone thinks is hard but is actually easy?”) gets a different but equally honest answer. Either reversal works because both invite the person to push back against the consensus, which is much more fun than confirming it.
2. Concession questions
“What's something you've completely changed your mind on?” Maybe the strongest single question in conversation. It signals you're willing to hear about ideological inconsistency, which most people are too polite to ask about, which means everyone has a stockpile of unaired material. Bonus: the answer reveals more about who someone is now than a hundred questions about their current beliefs could. The shape of how someone changes their mind is their character.
3. Constraint questions
“If you could only own five books for the rest of your life, what makes the cut?” The constraint forces a real answer. Without the constraint, the answer is “oh I love so many books” which is socially acceptable and totally useless. With the constraint, they have to actually think. The thinking is the conversation.
The trick is the size of the constraint. Five is the magic number for most domains. One is too few (forces them to caveat). Ten is too many (lets them be polite to all their favorites). Five forces a confession.
4. Texture questions
“Describe your perfect Tuesday afternoon in three sentences.” The day of week matters; Tuesday afternoon is mundane in a way Sunday morning isn't. The word “perfect” lets people answer with mild things instead of grand ones. The three-sentence limit kicks them past the surface answer into the actual texture.
Texture questions are how you find out what someone's life would feel like to be inside. They're slow questions; people pause to answer them. The pause is good.
The 30 prompts
Reversals (8)
- What's something most people get wrong about [your job]?
- What's a piece of common advice that's completely backwards?
- What's something everyone treats as serious that you find funny?
- What's something everyone treats as a joke that you find serious?
- What's an industry secret you wish more people knew?
- What's a famous “winner” you think is overrated?
- What's a forgotten thing you think deserves a comeback?
- What does your generation get wrong about the previous one?
Concessions (8)
- What's something you've completely changed your mind on?
- What's a belief you used to defend that you now find embarrassing?
- What's something you've been quietly wrong about for years?
- What's a piece of advice you ignored that turned out to be right?
- What did you used to think made you cool that doesn't?
- What's a hill you used to die on that you'd now retreat from?
- What's a friend you wrote off who you misjudged?
- What did you want at 18 that you'd hate now?
Constraints (7)
- If you could only watch five movies forever, what makes the cut?
- If you could only live in three cities for life, which?
- If you could keep one possession from your childhood, what would it be?
- If you had to teach a one-hour masterclass tomorrow, on what?
- If you could erase one decade and replay it, would you?
- You can save one art form from extinction. Which?
- You have one year to be world-class at something. What do you pick?
Textures (7)
- Describe your perfect Tuesday afternoon in three sentences.
- Describe a smell that makes you feel like you're home.
- What's a sound you associate with the happiest version of yourself?
- Describe the room you'd work in for the rest of your life.
- What's a small ritual that defines your week?
- Describe the perfect moment of friendship you've had this year.
- What's the smallest thing that makes you feel rich?
The follow-up rules
The prompt is the door. The follow-ups are the conversation. Three rules for those:
- Always ask why before asking what next. If they say “I changed my mind on X,” ask why before you ask “and what about Y?” The why is where it gets honest.
- Don't fill silence. The 4-second pause after they answer is when the second answer arrives. The second answer is the real one. Most interviewers ruin this by jumping in.
- Reflect, then probe. “So you're saying X — is that because of Y?” gives them a chance to correct the framing. Half the time they'll say no, and you'll get a sharper answer.
Where to use these
First date. Coffee with a colleague you don't know. The first 10 minutes of a podcast. Family dinner when the conversation has stalled. They work everywhere because they're built around how humans actually want to be asked things, not around what social scripts demand.
And if you're trying to use these in a podcast format and you don't have anyone to use them with, that's a problem with a known solution. PodRandom has 300+ of these organized by category, plus a way to be matched with a stranger who's ready to answer right now. The whole product is built around the assumption that good questions plus willing strangers is enough.