Stop Treating First Conversations as a Personality Trait, Start Training Them on Video Chat

I have lost count of how many times I have heard a capable professional say some version of "I'm just not a people person." It is usually said lightly, almost as a joke, and it is almost always treated as settled, as though it were a fixed trait handed to you at birth along with your height.

I have never found that convincing, though it is worth separating two different claims here. Whether sociability is a fixed trait is a question the research below does not settle, and I am not going to pretend otherwise. What that research does show is narrower and more useful: our judgements about how these conversations actually go are reliably wrong.

So what people are often describing when they say that may be less a personality than an untrained skill, one that has been quietly decaying for years in an environment that offers almost no opportunity to practise it and no instrument to measure the loss.

The advice we give about this is mostly useless

Walk into any professional development session on communication and you will be handed information: frameworks for difficult conversations, scripts for handling objections, a template for cold outreach that someone's agency tested once.

None of that material is wrong, it is simply the wrong category of thing.

A first conversation with someone outside your existing network runs in real time, under uncertainty, with no opportunity to pause and consult the framework. It sits much closer to sight-reading music than to anything you can absorb from a slide deck. Skills of that type improve only through repetition under live conditions, and they fade once the repetitions stop.

The fading is the part nobody notices, because nothing in working life measures it. A sales director who has not opened a conversation with someone outside her own network in three years has lost real capability. She owns no instrument that reports it. She just knows she dreads the conference.

What the research actually found

In 2014, Nicholas Epley and Juliana Schroeder published nine experiments in the Journal of Experimental Psychology: General involving more than 800 commuters, most of them on Chicago trains and buses. Participants were assigned to commute as usual, to sit in deliberate solitude, or to open a conversation with the person beside them.

The people who opened a conversation reported the most positive commute. A separate group asked to predict the outcome beforehand had ranked that same condition as by far the least pleasant. Epley and Schroeder later replicated the work in the United Kingdom in partnership with the BBC, on public buses and trains, with comparable results, as Chicago Booth documents on its own research pages.

Hold onto that gap between the forecast and the result. It is the whole problem in one finding.

Then there is the work of Boothby, Cooney, Sandstrom and Clark, published in Psychological Science in 2018. Across five studies, they documented what they named the liking gap. In lab conversations between people who had just met, among dormitory residents getting acquainted, and in a personal development workshop, participants consistently underestimated how much their conversation partners liked them, and the gap persisted for months. Trained coders watching the recordings could see the liking being signalled perfectly clearly. The participants simply did not register it.

So we walk in expecting it to go badly, it goes well, and we come out still believing it went badly. What that describes is not a personality trait but a measurement error, and measurement errors are the kind of thing you can correct.

4 conditions that turn practice into progress

The research I return to most often here is Anders Ericsson's work on expert performance, developed with Ralf Krampe and Clemens Tesch-Römer and published in Psychological Review in 1993 (volume 100, issue 3, pages 363 to 406). Their argument was that hours logged predict skill poorly. Improvement depends on a particular kind of practice, and that kind of practice has identifiable conditions, four of which matter here.

One caveat belongs here rather than in a footnote. A 2019 replication by Macnamara and Maitra did not reproduce the original violin finding at full strength, and reported a considerably smaller effect. The four conditions below still hold up as a description of what useful practice looks like, but the original paper's effect size does not.

1. A defined goal for each attempt, rather than general exposure.

2. Sustained attention while the attempt is happening.

3. Feedback fast enough to attach to whatever produced it.

4. Enough repetitions, with adjustment between them, for the adjustment to compound.

Now hold those four up against what a working adult actually gets.

Condition	What working life offers	What is missing
Defined goal	Networking events with a commercial target	The commercial target distorts the very behaviour you are training
Sustained attention	Conversations held while scanning the room	Attention is divided by design
Fast feedback	Politeness from colleagues and clients	Hierarchy and self-interest filter the signal
Enough repetitions	A handful of first conversations per quarter	Nowhere near the volume needed to adjust

Volume is the one that breaks everything. A violinist working toward competence accumulates thousands of hours. A professional trying to get better at first conversations gets maybe forty real attempts in a year, and every one of them carries enough consequence to discourage any experiment at all.

Between the low volume, the high stakes and the filtered feedback, you could not design a worse training environment on purpose.

Where the repetitions are actually available now

Something shifted in the last few years and professional development has not caught up with it. A large population now treats video conversation as an ordinary way to get to know someone outside their existing circle, rather than as a last resort when travel falls through.

Platforms built around shared language and shared interests let a person choose who to talk with, one at a time, for a set length of time. A twenty-minute video chat with someone outside your professional circle carries no career cost, no shared history, and no obligation to continue.

Measured against those four conditions, the format lines up reasonably well as somewhere to practise. Volume is effectively unlimited, and the stakes stay low enough that you can deliberately try something that might not work. Attention is undivided, because there is no room to scan. And a one-to-one format removes the group you would otherwise be performing for, which in my experience is the variable that most often gets in the way of a practice conversation.

Feedback is the condition that still takes deliberate effort. The liking gap tells us exactly why, and it also tells us what to do about it.

7 steps I give people who want to work on this

Set it up:

1. Pick one objective per session. Not "get better at conversation." Something you can actually score, such as asking three follow-up questions before you offer an opinion, or letting two silences run without filling them.

   
   

2. Decide your frequency and put it in the calendar. Three sessions a week for a month gives you around twelve repetitions, which is enough for a pattern to start becoming visible.

Run it:

3. Cap the session at twenty minutes. Camera conversations tire people faster than in-person ones do. Stopping while you both still have energy is what makes the next one happen.

   
   

4. Start with a question, not a summary of yourself. Most people open by establishing credentials. It is the single most common way to make the first two minutes feel like an interview.

   
   

5. Let the silences sit. This is the one that feels worst and pays best. Count to three before filling a pause and watch how often the other person fills it with something better than you had.

Keep it honest:

6. Afterwards, write down one observable fact, not a feeling. Did they ask a question back? Did the conversation run past the point either of you planned? Observable facts survive the liking gap in a way that "I think that went badly" does not, and you now know that verdict is probably wrong anyway.

   
   

7. Review the log monthly, not daily. The improvement in a skill like this is invisible across three sessions and obvious across thirty.

What this will not fix

I want to be straight about the limits, because overselling this would be easy. This trains one component only, which is the opening and sustaining of a conversation with someone who carries no prior context about you.

It will not train negotiation under pressure. It will not help with conflict involving a colleague who has standing to damage your career. It will not teach you to persuade a buyer who has a budget and a competing quote on the desk. Those are separate skills and they need their own repetitions in their own environments.

What it does address is the forecasting error Epley and Schroeder documented, which the liking gap then compounds and preserves. Both findings point in the same direction. The barrier to starting a conversation is usually a bad prediction rather than a shortage of charm.

Predictions improve with data, data requires attempts, and attempts are the one thing that has finally become available in volume.

So the next time you catch yourself saying you are just not a people person, try treating it as a training log problem instead of a personality one. Mine currently runs at twelve sessions a month. What would yours look like if you actually kept one?