Skip to main content

Communication

The Science of Great Conversations

Great conversation is not charisma. It is not extroversion. It is a small set of behaviors that have been measured, replicated, and ranked by researchers from Cornell, Harvard, and the University of Chicago — and almost everyone gets them wrong in the same predictable ways.

Why Most Conversations Fall Short

Picture a software engineer named Maya who has just left a holiday party. On the drive home, she replays the night in her head: that pause before she answered her boss’s question, the joke that did not land, the way her neighbor looked at her phone mid-sentence. By the time she pulls into her driveway, she has decided the night went badly and that her boss probably thinks she is awkward.

Cornell psychologist Erica Boothby has run this exact experiment — five times, with 622 participants (Boothby, Cooney, Sandstrom & Clark, Psychological Science, 2018). After every conversation, partners rated each other and predicted how much the other person liked them. In every study, the prediction came in below the actual rating. Boothby called it the Liking Gap, and it shows up in lab chats, in dorm rooms tracked for months, and in workshops up to 45 minutes long.

Maya’s boss almost certainly liked her more than she thought. So did her neighbor. So did the woman in the kitchen.

Bad conversations rarely happen because someone is bad at conversation. They happen because of three predictable cognitive errors the research has now mapped in detail:

Each of those errors is fixable with one small, learnable behavior change. The next sections show which ones.

Try this on the way home from your next event: when the cringe replay starts, name it out loud — “That’s the Liking Gap” — and assume the other person liked you 20% more than your inner critic is telling you. Boothby’s data says you will almost always be closer to the truth than the replay is.

Three Studies That Rewrote What We Know About Conversation

Most advice on conversation is folk wisdom: “be yourself,” “be a good listener,” “don’t overthink it.” Three peer-reviewed papers have replaced almost everything that wisdom got wrong. Together they form the spine of every recommendation in this guide.

1. The Liking Gap — people like you more than you think

Boothby and her co-authors ran five studies (622 total participants), pairing strangers for short conversations and tracking dorm-mates over months. After every conversation they asked two questions. How much did you like your partner? And: How much do you think your partner liked you? In every study, the second number came in lower than the first.

The gap was widest immediately after a conversation and shrank over time, but in the dorm-mate study it was still measurable months later. People who scored as more shy felt the gap more strongly than people who did not.

Our studies suggest that after people have conversations, they are liked more than they know.

What this means for everyday conversation: the post-event cringe spiral is, statistically, a cognitive bias — not an accurate read of the room. Try this: after the next conversation that makes you wince, write down what your partner actually said and did during the talk. Stick to observable behavior. The data almost always points the opposite direction from the spiral.

2. The Follow-Up Question Effect — curiosity beats charisma

Harvard Business School professor Alison Wood Brooks and her co-authors ran one of the largest studies of conversation behavior ever conducted: thousands of online conversations and a 106-dyad speed-dating field experiment, all coded with machine learning and trained human raters. They sorted every question into one of six types — introductory, mirror, full-switch, partial-switch, follow-up, and rhetorical.

The headline finding made the cover of Harvard Business Review: people who asked more questions were rated as more likable, and the active ingredient was follow-up questions — the kind that build directly on whatever a partner just said. Follow-ups made up about 44% of all questions and were the single strongest predictor of liking.

In the speed-dating arm of the study, the result was almost cinematic: daters in the top tercile of question-asking got a “yes” to a second date 39% of the time, versus 22% in the bottom tercile. Their average liking rating was 5.79 vs. 5.31 on a 7-point scale (Huang, Yeomans, Brooks, Minson & Gino, Journal of Personality and Social Psychology, 2017).

Compared to those who do not ask many questions, people who do are better liked and learn more information from their conversation partners.

The kicker: in Brooks’s research, most people do not realize asking questions makes them more likable. They believe they need to be impressive, witty, or interesting. They actually need to be curious.

Try this in your next conversation: for one full hour, replace every “that’s cool” with a follow-up question that builds on the last sentence. “What made you start that?” “What surprised you about it?” “What changed your mind?” People will tell you their best stories, and they will leave the conversation thinking better of you.

3. The Deep Conversation Paradox — depth feels better than people predict

University of Chicago behavioral scientists Nicholas Epley, Michael Kardas, and Amit Kumar ran twelve experiments with more than 1,800 participants. They paired strangers and asked them to discuss either shallow questions (“What’s the best TV show you’ve watched lately?”) or deep ones (“For what in your life do you feel most grateful?”). Beforehand, participants predicted that the deep conversations would feel awkward and unpleasant. Afterward, they rated those same conversations as significantly more enjoyable, connecting, and meaningful than the small-talk control (Kardas, Kumar & Epley, Journal of Personality and Social Psychology, 2022).

The miscalibration was extraordinary. In Experiment 2, the gap between predicted and actual awkwardness measured Cohen’s d = 2.23 — one of the largest effect sizes in the entire conversation literature. (Most “big” effects in social psychology run d = 0.5 to 0.8.) When researchers simply told participants about this miscalibration, those participants chose deeper topics for subsequent conversations on their own.

People seemed to imagine that revealing something meaningful or important about themselves in conversation would be met with blank stares and silence, only to find this wasn’t true in the actual conversation.

Try this: in one conversation this week, swap your default question (“What do you do?”) for a deeper one: “What’s something you’re working on that you actually care about?” The Kardas/Epley data says it will feel less awkward than imagined and more connecting than expected. Every time.

Two people leaning in across a cafe table, fully engaged in deep conversation.

What Actually Makes a Conversation Great

The three studies above explain why bad conversations happen. The next layer of research explains what good ones look like up close. The same five behaviors keep showing up in study after study:

  • Trade turns. Tickle-Degnen and Rosenthal’s foundational rapport framework (Psychological Inquiry, 1990) calls this mutual attentiveness: one of three components of any rapport-rich interaction, alongside positivity and nonverbal body language coordination. Hogging the floor breaks it. So does hiding from it.
  • Match the mode. Pulitzer Prize–winning journalist Charles Duhigg synthesized two decades of communication research in Supercommunicators (2024) and identified three conversation types: Practical (“What’s this really about?”), Emotional (“How do we feel?”), and Social (“Who are we to each other?”). Most miscommunication is not about content. It is about two people being in different conversations at the same time.
  • Lead with warmth, then competence. Susan Fiske’s stereotype-content research and the Captivate framework both point to the same equation: Warmth × Competence = Charisma. Pure cleverness without warmth reads as cold. Warmth without competence reads as nice but forgettable.
  • Listen like it matters. Guy Itzchakov’s program of research shows that high-quality active listening, attentive, empathic, and non-judgmental, measurably reduces a speaker’s anxiety, increases their attitude clarity, and even softens extreme attitudes during disagreement (Itzchakov et al., Human Communication Research).
  • Go one level deeper than feels comfortable. The Kardas/Epley paradox above, in one sentence: the move from “What do you do?” to “What’s something you’re working on that you care about?” shifts a conversation from forgettable to memorable every time.

Imagine an HR manager named Daniel walking into a coffee meeting with a senior leader he has been told to impress. Before reading the research above, he prepares three “interesting facts about himself” the night before. After reading it, he prepares three follow-up questions he might ask about whatever the leader brings up first. Same meeting. Different result. Brooks’s data says the second version of Daniel walks out more liked.

Diagram contrasting warmth and competence as the two dimensions of charismatic communication.

Pick one of the five behaviors above to focus on for a full week. Trying to install all five at once is how new habits die. Trying to install one is how they stick.

The Conversations Worth Stopping to Have

Three findings worth filing under “things people are systematically wrong about”:

Conversations almost never end when both people want them to

Adam Mastroianni and colleagues at Harvard analyzed 932 dyadic conversations and asked both participants when they would have preferred to wrap up. The two answers matched in fewer than 2% of conversations (Mastroianni, Gilbert, Cooney & Wilson, PNAS, 2021). About 48% ran too long. About 34% ended too soon. Almost everyone, almost all the time, misjudges the exit.

What this means: the awkwardness of “Should I leave? Do they want me to leave?” is not a personal failing. It is a coordination problem every human has. Use this script next time: “I have to run, but this was great — let’s pick this up.” It is short, it is warm, and it is almost always received better than the white-knuckle alternative of staying ten minutes longer than either person wants.

Talking to strangers makes commutes happier — but no one predicts it

Across more than 800 commuter and waiting-room participants, Chicago’s Nicholas Epley and Juliana Schroeder ran a counterintuitive experiment: they assigned some people to strike up a conversation with a stranger and others to keep to themselves. The “talk” group reported a more positive ride than the “solitude” group (Epley & Schroeder, Journal of Experimental Psychology: General, 2014). Beforehand, almost everyone predicted the opposite.

Try this on the next train, plane, or rideshare: ask one open question — “What brings you out today?” The Epley/Schroeder data says the upside is bigger than the prediction.

Weak ties matter more than weak ties feel like they should

Sussex psychologist Gillian Sandstrom’s research on weak ties — brief interactions with acquaintances or strangers — shows that the more weak-tie chats people have on a given day, the happier and more connected they report feeling. In a barista experiment, treating service workers with eye contact, a smile, and brief conversation produced higher positive mood than efficient, head-down exchanges (Sandstrom & Dunn, PSPB, 2014).

Picture a remote worker named Priya whose calendar is wall-to-wall meetings and whose only “in-person” interactions are quick coffee runs. Sandstrom’s research says the 30-second chat with the barista is not throwaway. It is a first impression micro-interaction that keeps her week from feeling sealed off.

Try this tomorrow: in one transactional interaction (barista, Uber, building lobby) make 10 seconds of warm small talk before the script kicks in. Notice how the rest of the day feels.

A Framework: The Three-Level Conversation Blueprint

The research above explains what makes conversation work. The natural follow-up question is how — what does a person actually do, in order, when they are face to face with another human?

That is the question Vanessa Van Edwards’s upcoming book, Conversation (Portfolio / Penguin Random House, October 2026), is built around. It synthesizes the studies on this page into a single repeatable structure: three levels every meaningful relationship moves through, each with three power questions.

Level Goal The Research It Operationalizes
Level 1 — Put People at Ease Safety + warmth in the first 10 seconds Tickle-Degnen’s positivity + attentiveness components; Sandstrom’s weak-tie research
Level 2 — Connect Deeply Curiosity about goals, worries, and values Brooks’s follow-up effect; Itzchakov’s high-quality listening
Level 3 — Create Meaning Shared identity and the stories people tell about themselves Aron’s “Fast Friends” paradigm (Aron et al., 1997, PSPB); Kardas/Epley’s deep-conversation paradox

Each level has three power questions. Together, all nine make up what Conversation calls the Power Question Set — the questions that consistently move people from polite small talk to “we have to do this again.”

The full blueprint, all 9 power questions, and the personality-decoding playbook that goes with them are available to preorder now (hardcover, ebook, audiobook).

Five Micro-Skills That Show Up in Every Great Conversation

Frameworks are the macro view. At the micro level, the literature converges on a handful of moves that take seconds to learn and a lifetime to use:

1. Conversational threading

When someone says, “I bought tires at the Goodyear station with my wife on Saturday,” the threads are: tires, Goodyear, wife, Saturday. Threading is the practice of pulling out the hooks in a partner’s last reply and using one to extend the dialogue — instead of giving a closed answer that forces them to start a new topic. It is especially effective for introverts because it spreads the conversational load.

Action: count the proper nouns in the last sentence your partner said. Ask about one of them.

2. Looping for understanding

Duhigg’s signature move from Supercommunicators: ask a question that confirms what someone said → repeat it back in different words → ask if you got it right. It sounds basic. In practice, it is the single biggest reason a conversation feels like it lands.

Action: in your next emotional conversation, replace the first solution you think of with a loop: “So what I’m hearing is _____. Did I get that right?”

3. Subtle (not performative) mirroring

Yale researchers Tanya Chartrand and John Bargh’s chameleon-effect studies (1999, JPSP) showed that people automatically mimic each other’s posture and mannerisms — face-touching went up 20%, foot-wagging 50% — and that being mimicked makes interactions feel smoother and partners feel more likable. The catch: the effect works because it is nonconscious. Performative mirroring backfires.

Action: match energy, not gestures. When the other person leans in, lean in. When they slow down, slow down.

4. Validation, not agreement

A core counseling micro-skill that ports cleanly into everyday conversation: “That makes sense.” “That frustration is valid.” “Of course you’d land there.” No one has to agree to make someone feel heard, and feeling heard is the empathy signal that makes them open up further.

Action: before disagreeing in your next hard conversation, validate first. The disagreement lands twice as well.

5. Named curiosity

Saying out loud what you are curious about: “I’m curious what made you pivot.” “I’m wondering how that landed at home.” It is a form of structured inquiry, and it does double duty — it asks a follow-up question (Brooks) and signals high-quality listening (Itzchakov) in the same beat.

Action: in your next conversation, name your curiosity once. Watch how it changes the depth of the answer.

A young woman listening intently with focused, curious attention.

Six Conversation Myths the Research Quietly Debunks

Myth What the research actually says
“93% of communication is nonverbal” (Mehrabian’s 7-38-55 rule) UCLA’s Albert Mehrabian’s 1971 work applies only to messages where words and tone conflict about feelings or attitudes — not to all communication. The “rule” is one of the most over-applied stats in the field. (Big Think)
“Extroverts are better conversationalists” No correlation between talkativeness and conversational insight. Susan Cain’s research on introversion and reams of follow-up data show introverts often listen more, follow up better, and report deeper conversations. (MIT Sloan Review)
“You should mirror exactly” Over-mimicry backfires. Chartrand and Bargh’s chameleon effect works because it is subtle; performative mirroring reads as creepy or sycophantic. (Chartrand & Bargh, 1999)
“Small talk is meaningless” Sandstrom’s weak-tie research shows brief small-talk interactions reliably boost belonging and happiness. University of Arizona researcher Matthias Mehl’s follow-up to his own “Eavesdropping on Happiness” study found small talk is neutral, not harmful — substantive talk is the additional ingredient, not a replacement.
“Just ask open-ended questions” Open-endedness alone is weak. The active ingredient in Brooks’s data is follow-ups — questions that build on whatever a partner just said. (Huang et al., 2017)
“Strangers don’t want to talk to you” Epley and Schroeder’s commuter studies: people assigned to talk to a stranger reported significantly more positive commutes than those assigned to keep to themselves — and predicted the opposite beforehand. (Epley & Schroeder, 2014)

Where to Start in the Next 24 Hours

The research is the easy part. Translation into one’s own life is the harder part. These are the SOP team’s most-used companion guides — pick one and run with it today.

Conversation Games

27 games that practice the same behaviors the research keeps pointing to — turn-taking, escalating depth, shared attention.

Read More

Frequently Asked Questions

What does the science actually say makes a great conversation?

The most replicated findings come down to four behaviors: asking follow-up questions, going one level deeper than feels comfortable, matching a partner’s conversational mode (Practical, Emotional, or Social), and offering high-quality listening. Studies from Harvard Business School (Brooks et al., 2017), Cornell (Boothby et al., 2018), and the University of Chicago (Kardas et al., 2022) all converge on this short list.

Why do my conversations feel like they went worse than they actually did?

That is the Liking Gap, documented by Cornell’s Erica Boothby and colleagues in Psychological Science (2018). Across five studies and 622 participants, people systematically underestimated how much their conversation partners liked them. The effect was strongest for shyer participants. The post-conversation cringe spiral is, statistically, a cognitive bias — not an accurate read of the room.

Are deep conversations with strangers really better than small talk?

According to research from Nicholas Epley’s lab at the University of Chicago — yes, and by a wide margin. Across 12 experiments with more than 1,800 participants, people predicted deep conversations with strangers would feel awkward but rated them as significantly more enjoyable and connecting than the small-talk control. The effect size on the awkwardness gap measured Cohen’s d = 2.23 — one of the largest in the literature.

What is the single highest-leverage thing I can change about my conversations?

Asking more follow-up questions. In Alison Wood Brooks’s speed-dating study, daters in the top tercile of question-asking earned second dates 39% of the time — versus 22% in the bottom tercile. Follow-up questions specifically — the kind that build directly on whatever a partner just said — were the strongest predictor of liking. The behavior is small, learnable, and almost no one realizes how much it works.

What are the three types of conversations?

Charles Duhigg, drawing on decades of communication research, argues that every conversation is one of three types. Practical conversations are about solving problems and exchanging information. Emotional conversations are about feelings and validation. Social conversations are about identity and belonging. Most miscommunication happens when two people are in different types at the same time without realizing it — one came in for emotional support, the other went straight to fix-it mode. Same words; different conversations.

How do I know when to end a conversation?

Probably nobody knows — and that is normal. Adam Mastroianni and colleagues analyzed 932 dyadic conversations and found fewer than 2% ended at a moment both partners wanted (PNAS, 2021). About 48% ran too long, 34% ended too soon. The implication: a graceful close — “I have to run, but this was great, let’s pick this up next time” — is almost always better received than the white-knuckle alternative of staying.

How do you connect with anyone, even if you're shy or introverted?

This is exactly what Vanessa Van Edwards teaches in Conversation: lead with genuine curiosity, ask follow-up questions, and go one level deeper than feels comfortable. Shy and introverted people consistently underestimate how much others already like them, so connection is usually closer than it feels. Her three-level Conversation Blueprint turns these into learnable moves anyone can practice.

How do you make a conversation memorable?

Skip the small-talk script and ask one question that invites a real answer. In Conversation, Vanessa Van Edwards calls these the power questions — three for putting people at ease, three for connecting deeply, and three for creating meaning. People remember feeling genuinely seen, and one sincere power question does that more reliably than any impressive story.

Where can I learn the full system?

The most current synthesis is Conversation: How to Connect with Anyone & Make Every Interaction Count by Vanessa Van Edwards (Portfolio / Penguin Random House, October 2026). It pulls the peer-reviewed research on this page into a three-level Conversation Blueprint with 9 power questions — three for putting people at ease, three for connecting deeply, and three for creating lasting meaning. For complementary reads, see Captivate (2017) on first impressions, Cues (2022) on charismatic communication, and the SOP team’s roundup of the 11 best books on conversation skills.

People School 10,000+ students

After People School, Debbie got a $100K raise. Bella landed a role created just for her.

The science-backed training that turns people skills into career results. 12 modules. Live coaching. A community of high-performers.

Explore People School Take the Free Quiz
Debbie
Bella
Kristian
Stacy
Join 10,000+ students · Results vary