Stop Asking 'What Do You Do?': A Field Guide to Conversations That Actually Connect

Across speed-networking experiments with 500 people, the most common opener in professional life, 'What do you do?', came dead last, while 'What was the highlight of your day?' beat every other question tested. That gap points to something most people never learn: conversations have depth levels, and the questions that work at one level fail at another. This guide gives you the map, the mechanisms behind it and the exact moves: how to open, how to go deeper without overstepping, and how to read the story a person tells about their own life.

Why conversation is a skill worth training now

The WHO Commission on Social Connection estimates that about 1 in 6 people globally, or 15.8%, report feeling lonely. Its estimates tie loneliness to 871,000 deaths each year over the period 2014 to 2019. Young people report the highest rates, with 21% saying they are lonely, and loneliness reaches 24.3% in low-income countries against 11% in high-income ones.

The same thinning shows up in daily life: only three in 10 Americans socialize on a day-to-day basis, according to a figure cited in the source. Social disconnection is also linked to heart disease, stroke, depression and anxiety, so the stakes run well beyond comfort.

Part of the erosion is technological: people now talk to technology more than to each other. Voice notes are a telling case, because a voice note is a monologue, which the source calls the most selfish form of conversation. Researchers studying large sets of real conversations break turn-taking into three parts: how the floor passes from one speaker to the next, how long each speaker holds it, and the small signals listeners send while the other talks, such as nods or a quick 'mhm'.

The flip side is opportunity: when fewer people practise, those who are masterful at conversation gain an advantage. The source frames conversation as the vehicle to get what you want, whether that is a promotion, a partner or friendships that hold loneliness at bay. It also insists that the skill is learned, not inherited: in its words, 'the one skill that keeps on giving'. Experiments back the learnability claim: when people were instructed to ask more questions, partners perceived them as more responsive, a quality that covers listening, understanding, validation and care. In one chat study by Harvard researchers, 430 participants were paired so that one partner had to ask at least nine questions and the other at most four, and asking more questions went with being liked more.

Conversation has become a research field of its own, with researchers treating it as one of the most complex human tasks and distilling it into rules such as the TALK maxims: Topics, Asking, Levity and Kindness. Another popular framework holds that every conversation is really one of three, practical, emotional or social, and that connection fails when two people are having different ones. The method at the heart of this guide comes from Vanessa Van Edwards' book 'Conversation, how to connect with anyone and make every interaction count'. The host describes it as 20 years in the making, and its systematic research began eight years ago with a series of speed-networking experiments. What sets it apart is a map of depth: it tells you not only what to ask but when each kind of question belongs, with the stated aim of making you feel more in control and more confident.

The map: McAdams' three levels of knowing a person

McAdams' three-level model of personality, adapted by Van Edwards into a map of conversational depth.

The backbone of the method is a framework from personality psychology built by Dan McAdams, which Van Edwards read while doing her conversation research. McAdams described persons on three relatively independent levels. At level one sit dispositional traits, like those of the Big Five, which give a general, comparative and unconditional sketch of someone. At level two sit tasks, goals, projects, values and other concerns that place a life in a time, a place and a role. At level three sit life stories: internalized narrations that integrate a person's past, present and imagined future.

McAdams' own image makes the layers easy to remember: traits sketch an outline, the concerns of level two fill in the details, and the life story speaks to what a life means overall.

Van Edwards turned this description of personality into a description of conversation, citing McAdams' finding that conversation is not as fluid as we would like to think. In her version, level one is called general traits, level two is personal concerns, where people begin to connect deeply, and level three is the self-narrative, the story you tell yourself about yourself. Each level calls for its own kind of question, and the sixth question of her set is a signature question that each person writes for themselves.

The host's own question, what the difference between level one and level two is, is the point where most readers get stuck. A question pitched at level three is too advanced for someone who is introverted, and skipping level one does damage to the potential connection later.

McAdams also argued that there is no reason to expect strong consistency across the levels, because lives are complex and often contradictory.

Level one: turn facts into stories

The trap of level one is getting stuck in facts. The source's rule for escaping it is worth memorizing: 'a fact is a stop sign, a story is a doorway.'

The strongest evidence in the method comes from three speed-networking experiments with 500 speed networkers. The best conversation starter among those tested was 'What was the highlight of your day?' 'What do you do?' ranked as the worst of the set.

Level one is not a hurdle to clear quickly, because skipping it damages the connection later. Its job is to create ease, which the source calls another way of saying warmth.

Searching for good is not the same as forcing cheer: toxic positivity can kill a conversational vibe very quickly. For people who find feelings questions hard, a numbers question is a gentle bridge: if a couple says they have been together 38 years, ask what the highlight of those 38 years has been. It works well for introverts because it is fact-based, built on something they already told you.

Boring questions make you boring, and the source names 'Been busy?' the worst question in the history of conversation. Example: 'Fine, just busy.' 'Busy sounds like an understatement. Busy with what, the good kind or the draining kind?'

Because the brain wants the simplest, easiest road possible, the answer should give a clear label first and the interesting part second. The source's formula is 'I do blank for who by how'. Career coaches converge on close cousins: 'I help who achieve what by how', or a three-step 'You know how...? So what I do is... In fact...' that opens on a shared problem and ends on proof. A version written for non-native speakers keeps it plainest: what I do, who I do it for and the result I create, followed by a short question back such as 'How about you?'

Template to adapt: 'I'm a [plain job title]. I [what you do] for [who], by [the unusual how]. What about you?' Example: 'I'm an accountant for bakeries; I'm the person who tells them which pastry is secretly losing money.'

Howard Stern, however one feels about him, is offered as a master conversationalist, and an analysis of thousands of his interview transcripts found one pattern: he creates information gaps. Curiosity research describes the mechanism: curiosity is a form of deprivation that arises when we perceive a gap in our knowledge or understanding. Template to adapt: 'What's the most surprising thing about your job that outsiders never guess?'

Level two: build dreams, never kill them

Level two is personal concerns, where people begin to connect deeply. Its purpose is to make someone feel heard and understood.

The central skill here is what the source calls dream building. Done well, it gives the other person a quiet motivation: in the conversation, they discover an aspect of themselves. Research on question-asking points to the same lever: follow-up questions are what make a partner feel responded to, and that responsiveness is what drives liking.

Ask about someone's pet project, the thing they tend to like a pet at home. Steer away from pest projects, the ones that weigh on a person, because pest projects carry a lot of shame. 'What is your biggest goal right now?' is a strong question, and it gets stronger when you anchor it in time.

Van Edwards adapts the move to her own setting: after speaking or a book signing, she asks 'what was your favorite part? Or what did you learn the most?'

The opposite of dream building is dream killing. Clever people often do it by accident, answering a dream with logic: they are fact checking, not gut checking.

Competence without warmth is likely to leave us feeling suspicious. Social cognition research explains why the order matters: people first judge whether someone intends good or ill, and only then whether that person has the ability to act on those intentions. Generosity scales: in one finding the source cites, the most popular kids were those with the longest list of people they liked.

People remember the first few minutes and the last few minutes of a conversation, so the exit deserves as much care as the opener. A meta-analysis of 174 effect sizes found the peak-end rule large and robust: evaluations of a past experience lean heavily on its most intense moment and on its end. The source's exit is a future Q, asking about or mentioning something coming up, plus a compliment or an action.

🛠 How to — Run one conversation from opener to exit

1. Open with a question that asks for a story, such as 'What was the highlight of your day?', instead of 'What do you do?'.
2. Once the exchange feels easy, bridge to level two by asking about a pet project or the person's biggest goal right now, anchored to a moment in time.
3. Build the dream: ask what it would look like, what draws them to it and what comes next, and keep every feasibility objection to yourself unless invited.
4. Keep your own turns between 15 and 30 seconds, and treat 45 seconds as the ceiling.
5. Close with a future Q about something coming up for them, then add a compliment or an action such as an introduction or a follow-up.

Level three: open the door slowly, and together

Going there too fast has a name, vulnerability regret: you shared something too vulnerable, and it was not a safe space or it was not received well. The source draws the line between authenticity and speed: being authentic is great, but being too vulnerable too quickly is very awkward and sometimes not safe.

The host, drawing on almost 700 interviews, holds the opposite intuition: 'vulnerability is the doorway to connection'. The first is escalation: in the best-known closeness procedure, pairs spend 45 minutes on self-disclosure tasks that gradually escalate in intensity, and they end up closer than pairs doing comparable small talk. The second is reciprocity: pairs who took turns disclosing felt more attraction than pairs in which one person disclosed while the other listened, and responsiveness, enjoyment and the sense of being liked explained the difference.

Let's call it the one-rung rule: share one step deeper than the other person just did, never several, and wait for them to match before you go further.

Keep the liking gap in mind: after conversations, people systematically underestimate how much their partners liked them and enjoyed their company.

Self-narratives: redemption, contamination and the luck question

Your self-narrative is the story you tell yourself about yourself, the way you make sense of your journey and purpose in life. Psychologists call it narrative identity: an internalized, evolving life story that joins the reconstructed past and the imagined future to give a life some unity and purpose. The source sorts these stories into two types: contamination narratives and redemption narratives.

In a study of midlife adults and college students, redemption sequences went with higher psychological well-being, while contamination sequences predicted low well-being among the midlife adults. A later review found that people whose stories feature redemptive meanings, personal agency and exploration tend to enjoy higher mental health, well-being and maturity.

The source goes further: people with redemption narratives, it says, live longer, have better immune systems, have better connections and feel happier. Read those links as associations, not guarantees: the field itself lists disentangling cause and effect between life stories and adaptation as a task for future research. A redemptive frame is a healthy habit of mind, but it is no substitute for medical or psychological care when someone is struggling.

Underneath the two frames sits a well-studied psychological variable, the locus of control. In Australian panel data, people with an internal locus reported higher life satisfaction and better mental health, partly through more physical activity and more social interaction. A laboratory experiment tied an internal locus to a preference for agency, an effect driven by women, and found no evidence that it worked through optimism or an illusion of control.

The locus is not fixed: after major financial problems or a serious illness, adults became 0.1 to 0.3 standard deviations more external on average, and some effects persisted 4 or more years later.

The most revealing narrative question in the method sounds like small talk: 'Do you believe in luck? Do you believe you're lucky? Do you believe in fate?' If someone says yes right away, that tells you a lot about them: it signals gratitude, agency and confidence.

The research behind the question comes from the psychologist the source calls Richard Weisman, who built a 10-question luck diagnostic. He concluded that, to a large extent, people make their own good and bad fortune. He describes four principles lucky people use: they maximise chance opportunities, listen to lucky hunches, expect good fortune and turn bad luck to good. His 'Luck School' exercises taught participants to think and behave like lucky people, and a significant number reported positive, lasting change in perceived luck, self-esteem, confidence and health.

Use it as a doorway, not a test. Example: 'Yes, actually, I got this job because I missed a train and met my future boss on the platform', which invites the follow-up 'What made you start talking to a stranger?' Example: 'Not really, things never go my way', which calls for no argument, only a gentle 'What's one time something did go your way?' and the patience to hear it out.

The five self-narrative archetypes. Treat each as a working hypothesis, never a fixed label.

The five self-narrative archetypes, and how to respond to each

The source sorts those plots into five archetypes. Treat an archetype as a working hypothesis about someone's story, never as a label you hand them.

The first archetype is the ambitious hero: someone who is never quite satisfied and keeps chasing the next shiny idea. Bartlett placed himself here and quoted Jeff Bezos: 'I've got enough ideas to sink the company.'

The contented pragmatist, in her words, is 'somebody who likes it as it is.' The selfless helper builds a story around serving others, and Van Edwards explains why that generosity attracts takers by citing Bill Eddy, who estimates that 10% of the population are high-conflict people.

Creative disruptors 'do not like rules', and their whole ethos is breaking things; the book uses Edward Snowden as one example. The unlucky struggler carries a contamination narrative, the story in which good things keep turning bad, and Van Edwards illustrates it with George Costanza from Seinfeld. Here the rule of the luck question applies: meet a 'no, I'm not lucky' with curiosity, never with a correction.

Turn-length guide based on Van Edwards' norms and CANDOR corpus research. The 200 ms gap between speakers appears consistently across languages.

The source offers a playful door: 'What book, movie or TV character are you most like?' Bartlett chose The Pursuit of Happyness, because at 18 or 19 he was shoplifting food to feed himself and his rent was five months late. Van Edwards chose Claire Dunphy, and added that she identifies first as a mother, not an entrepreneur.

How these archetypes map onto the entertainer and appreciator styles is a question the source leaves open.

Talking less: turn length, overtalking and the fear of silence

The host cited a survey in which 95% of people had worked with someone who talks too much. Those chatty coworkers spend an average of 90 minutes of their workday just talking, and 71% of respondents said talkative colleagues kept them from getting their work done. The survey is not identified by publisher, year or method, so read those numbers as a sign of how common the irritation is, not as a precise measure.

The practical norm Van Edwards gives is a turn of 15 to 30 seconds; 45 seconds is acceptable, but beyond that a turn starts to become a monologue. She attributes the mechanics to the CANDOR corpus, a large research dataset on how conversation actually works. That corpus holds 1656 unscripted video-chat conversations, more than 7 million words and over 850 hours of audio and video. Its authors replicate a striking finding: the typical gap between one person stopping and the next starting is about 200 milliseconds, roughly an eyeblink, and it appears consistent across languages and cultures.

Listeners also send small signals while you talk, such as nods and short words like 'mhm', 'yeah' or 'exactly', which researchers call backchannel feedback.

The source gives two ways to measure yourself instead of guessing. The first, which she calls the yap challenge, is to record yourself giving an unscripted opinion on your phone camera and time it. The second is a five-item 'do you talk too much' test: if you answer 'often' more than three times, you are probably an overtalker. The threshold is phrased slightly differently at another point of the conversation, so read your score as a warning light, not a diagnosis.

Communication researchers studied this pattern under the name conversational narcissism: an extreme focus on the self in conversation, at the expense of appropriate concern for the other person. People recognise it through boasting, steering the topic back to themselves, exaggerated gestures, a loud voice and glazing over when others speak, and people enacting it were rated lower on social attraction. The same studies found contexts where focusing on yourself is appropriate, so the goal is balance, not silence.

Van Edwards adds a humane observation: some conversational narcissists are aware of it and hate themselves for it.

Entertainers and appreciators: read who is in front of you

Before you choose a question, know what kind of conversationalist you are facing. The source divides people into two types: entertainers, who talk more and want to be asked questions, and appreciators, who talk less and want to be told stories. This is her answer to a question many people carry into the room: how do I talk to a VIP, or to my boss? Let's call the rule stage or show: an entertainer wants a stage, an appreciator wants a show.

Template to adapt: with an appreciator boss, 'Something happened in the client meeting that I think you'll enjoy', followed by a short story; with an entertainer boss, 'How did you decide to take that project on?'

Underneath both styles sits one principle: the wrong way to approach conversation is trying to impress rather than to discover. Van Edwards credits Dale Carnegie with the idea that moved her toward less selfish conversation, usually summed up as being interested, not interesting. Carnegie's original line in How to Win Friends and Influence People is 'To be interesting, be interested', which makes interest the route to being interesting rather than its opposite.

The stereotype content model: warmth is judged first, competence second. Leading with warmth unlocks the admired quadrant.

The source also holds that extroverts can fall into conversational narcissism, while ambiverts and introverts often do very well. Research adds a useful nuance: extraverts do tend to build rapport more easily, but one study traced the advantage to mimicking their partner, and only when they were motivated to connect. A summary of that work puts it plainly: extraverts were more likable not because they talked more, but because they were better at mimicking other people's body language.

Bartlett observed that the people with aura in his life are not the most successful or accomplished, but the ones he would most like to spend time with. Van Edwards agreed that most of the billionaires she has met are quite humble, while admitting she has not met many.

Warmth first: the science of charisma, voice and eye contact

Charisma, rizz or aura, in Van Edwards' definition, is simply being someone others really want to be around. In the first moments, people answer two silent questions about you: can I trust you, and can I rely on you? The first is warmth and the second is competence, and she says 82% of our impressions of people are based on these two dimensions.

The research explains why the order matters: when we meet someone, we first judge whether they intend good or ill, and only then whether they can act on those intentions. People seen as both warm and competent draw uniformly positive reactions, while those high on one dimension and low on the other draw ambivalent ones. The stereotype content model links the high-competence, low-warmth combination to envy. That is the mechanism behind her warning that competence without warmth is likely to leave us feeling suspicious. Lead with warmth, then let competence show.

The second lever is to signal clearly that you like the other person, because people underestimate how much they are liked. Researchers named this the liking gap: after conversations, people systematically underestimated how much their partners liked them and enjoyed their company, from lab strangers to first-year college dorm mates. The cause is not missing signals: the signals are right there, but people estimate what others think of them from their own, overly critical view of themselves. Template to adapt: 'I really enjoyed hearing about your garden project, I'll be thinking about that hedge all week.'

Van Edwards calls it winner energy: some people walk into a room and have claimed their space. Bartlett named Michelle Obama and Tony Robbins as people with that quality. Her vocal tool is the maximum resonance point, the lowest end of your natural range, which she links to pride and calm. For her, that point comes from speaking on the out breath, with relaxed vocal cords.

The MIT Communication Lab's exercises start with diaphragm breathing: one hand on the belly, one on the back, feeling the gap between them widen as you breathe in. For resonance, say 'Mmmm' and feel the buzz at the front of your face, then carry it into words, as in 'Mmm-hmm one, Mmm-hmm two'. An NHS voice guide adds humming with a relaxed jaw and teeth apart, and notes that a yawn relaxes and opens the throat. Aim for your own low, relaxed range, not a forced deep voice: pushing below your natural floor strains the voice and sounds staged.

Eye contact is chemistry and computation at once: Van Edwards notes that it produces oxytocin, which helps connection, but limits how much you can process. Her guidance is to stay around 60 to 70% eye contact in Western cultures, not more. A cross-cultural study fits the processing point: Canadians held eye contact 64% of the time when they knew an answer, but 47% when they had to think it through. Norms vary widely: in the same study, Trinidadians held eye contact 88% of the time on known answers and Japanese participants 54%. A dual eye-tracking study also challenged the idea that East Asians avoid gaze, finding more mutual gaze in East Asian pairs than in Western pairs during live interaction. Van Edwards herself limits her norm to Western cultures, and notes that even nodding differs in places such as India, Bulgaria and Pakistan.

One listening cue is worth learning: the lower-lid flex, a slight tightening of the lower eyelids, which she calls a universal response. It is 'not positive or negative', but a cue of intense focus. When you see it, you have touched something that matters to the person: bookmark it to come back to, or explore it right away. Neil deGrasse Tyson reportedly built his books from every eyebrow-raising thing he told strangers on planes, the same principle applied to a whole career: faces show you what lands.

Exits that stick, and a question that is yours

The source grounds the exit in the peak-end effect: we remember the first few minutes and the last few minutes of a conversation. The research version is slightly different and more useful: a meta-analysis of 174 effect sizes found that people judge a past experience mainly by its most intense moment and its final moment, with a large effect of r = 0.581. The same analysis found the peak-end effect stronger than the effect of the beginning, and the effect of duration essentially nil.

Van Edwards' exit formula is a future Q plus a compliment or an action. A future Q means asking about or mentioning something coming up in the other person's future. The alternative is what she calls the worst kind of last impression: a lingering awkwardness.

Among the nine questions of her framework, the sixth is a signature question, one you write yourself. Amy Poehler ends her podcast by asking 'what are you laughing about?', and Jenny Yurick, behind the thousand-hours-outside challenge, asks people about their favourite outdoor memories. Van Edwards' own current signature question is 'what's your favorite conversation starter?' Example: a keen cook might use 'What's a meal you still think about?', which invites a story and a sensory memory in one go.

She suggests writing down the questions you are asked most often and crafting answers that are gifts to the other person, not just interesting to you. Her example is a friend, Jason, who cannot smell or taste and is always at dinners with people; he turns that into a gift by asking his companions what he should taste. Nicole Lappin remembered him through exactly that story and began telling it to Van Edwards, which is what a gift answer does: it travels without you.

Screens, voice notes and the office

Van Edwards calls the voice note 'the most selfish form of conversation', because it is a monologue. Responsiveness, enjoyment and the feeling of being liked carried that difference.

The stakes are not trivial: the World Health Organization lists excessive or harmful use of digital media, especially among younger people, among the risk factors for loneliness. On technology, Van Edwards is no purist: she sees AI as a good brainstorm partner and a good partner for practising conversation.

The sharpest debate in the conversation was about dating: is asking for someone's Instagram handle instead of their number a good move? Van Edwards countered that a feed front-loads level one: the other person arrives on a first date already holding a curated, performative version of you, which makes the date awkward. A first date exists to answer live questions: are you funny, will you ask me good questions, are you a conversational narcissist?

At work, the cost of getting this wrong is concrete: communication friction costs companies money. Her team intervention is small: once a week, at the start of a meeting, open with 'tell me something good'.

Example: a team of eight spends ten minutes on the round each Monday; a designer mentions her first half-marathon, and the following week two colleagues ask how training went.

A practice plan, and the mistakes to watch for

Let's call the plan below the ladder: each week climbs one stage of a conversation, from your own turns to your exits. Week one, measure: do the timed recording and the overtalking test before changing anything. Week two, openers: replace your default opener with a story-inviting one and note which stories come back. Week three, level two: ask one goal question per conversation and practise building on the answer; week four, exits: end every conversation you can with a future Q plus a compliment or action.

At the opening, it is the boring question; as Van Edwards puts it, boring questions make you boring. In the middle, it is the monologue and the logic-driven reply that kills a dream; at depth, it is sharing or asking too much too soon, which she calls vulnerability regret. At the end, it is drifting off with nothing said, leaving the lingering awkwardness she warns against.

Van Edwards frames the whole goal as feeling more in control and more confident, and that is the right measure of progress: less dread before the next room you walk into.

Worth quoting

• A good opener does not ask what someone is; it asks what happened to them.
• Facts close loops. Stories open them.

Key points

• Ask for stories, not facts: the highlight of someone's day beat every other opener tested, and 'What do you do?' came last.
• Climb the levels in order, traits, then personal concerns, then self-narrative, and gate the last one for safe relationships.
• At level two, build dreams with questions about goals and pet projects; logic and fact checking kill them.
• Keep your turns at 15 to 30 seconds, with 45 seconds as the ceiling.
• Endings weigh heavily in memory, so close with a future Q plus a compliment or an action.

Action plan

1. For every new conversation this week, replace your default opener with 'What was the highlight of your day?' and note which answers turn into stories.
2. Write your answer to 'What do you do?' with the 'I do blank for who by how' formula, then say it aloud until it takes under 15 seconds.
3. Record yourself giving an unscripted opinion on your phone camera, time it, and cut any habit that pushes past 45 seconds.
4. Write down the questions you are asked most often and craft answers that are gifts to the other person, not just interesting to you.
5. End every conversation this week with a future Q plus a compliment or an action.
6. Start one weekly team meeting with a round of 'tell me something good'.

Test yourself

Q. In McAdams' original research, what belongs to level two of describing a person?
A. Tasks, goals, projects, values, defenses and other motivational concerns that place a life in a time, a place and a role, sometimes called characteristic adaptations.

Q. What did research on redemption and contamination sequences find, and what is its limit?
A. Redemption sequences went with higher well-being and contamination sequences predicted low well-being in midlife adults; the field still has to disentangle cause from correlation.

Open questions

• Do the winning openers from the speed-networking experiments hold across cultures, ages and industries?
• How long a single turn should last is still poorly mapped, since turn duration remains understudied compared with turn exchange, partly for lack of large, good datasets.
• Are endings weighted because they come last, or because of other processes, given research finding that endings are not inherently over-weighted in evaluations?

Go further

1. Personality, Modernity, and the Storied Self: A Contemporary Framework for Studying Persons — McAdams' foundational article laying out the three levels of personality: traits, concerns and life stories.
2. When Bad Things Turn Good and Good Things Turn Bad: Sequences of Redemption and Contamination in Life Narrative and their Relation to Psychosocial Adaptation in Midlife Adults and in Students — The study that defines redemption and contamination sequences and links them to well-being in adults and students.
3. Luck – PROF Richard Wiseman — Wiseman's summary of his luck research, with the four principles lucky people use and the Luck School exercises.
4. The Liking Gap in Conversations: Do People Like Us More Than We Think? — The liking gap study showing people underestimate how much new conversation partners like them.
5. Asking Questions Increases Likability – Association for Psychological Science – APS — A readable account of the experiments showing that asking more follow-up questions increases likability.
6. Social connection — WHO's current figures on loneliness by age and country income, and its effects on health and work.
7. Talk: The Science of Conversation and the Art of Being Ourselves - Book - Faculty & Research - Harvard Business School — A research-based book on conversation built around the TALK maxims: Topics, Asking, Levity and Kindness.