How to have great conversations

There's a genre of funny skits popping up around the internet about conversations and how ChatGPT has ruined our ability to connect with others. We just get too nervous about what to say and have to refer back to the AI to get our next response.

There's something to this, but it's a problem that's been around a lot longer than AI has. Some people get nervous around others (and so confuse themselves with introverts) and don't seem to be able to connect on the level they'd like, if only they were able.

But you can learn to have good conversations, and the research shows that when you do, they become much more rewarding for both parties.

There are lots of component parts to conversations, but we can roughly split them up into three stages (based on the research of Alison Wood Brooks):The beginning

You can think of this as the foreplay of a conversation, the small talk which allows you to go more deeply into a relationship with the person you're speaking to. The purpose of conversation at this stage is merely to build trust, to build an initial connection to test for receptiveness, attention, care, warmth. There's nothing too profound at this stage of conversation, but you can lay the groundwork here for going deeper if the other person is in that mind space.

You can think of conversations as having many branches; alternatively you could think of them as hooks. These are opportunities that you and your conversational partner lay down together for exploring different topics or for going deeper into current topics. And paying attention to these as you go through the conversation will always provide you with ample material to talk about.

Small talk then, does a few things. It builds trust. It shows that you are likable and relatable and somebody that they want to have a conversation with. And it also embeds some of these hooks at the beginning so that the conversation can become richer more quickly.

Bad small talk might be, “hey, how are you?” Or “what's going on?” These are perfunctory questions, what Alison Wood Brooks calls mirror questions. They don't provide any useful data or conversational branches to continue down, and they waste a turn in the conversation.

An alternative would be to ask questions that center around values, which allow the other person to share something of themselves that gives you a hint into who they are. A few good questions in this vein might be, “what's exciting you at work lately,” or “what was it about where you lived that made you want to live there?” These questions allow you a deeper insight into what the other person values, and they show you a little bit about who they are and provide you with an opportunity to decide if you'd like to continue down that road, or if this is somebody actually you don't want to talk to.

Another point to keep in mind is that in any good conversation, there has to be a match of energies and depth between the two people speaking. The easiest way to think about this is in terms of vulnerability. You want to have enough vulnerability that the other person feels like they're learning about you, so not too little. Being too vulnerable is also dangerous though, because you put weight onto the conversational structure that it can't yet support, and the entire thing collapses. So you want to make sure that you're not being too overly vulnerable at the very beginning of a conversation in a way that doesn't match the context or the depth of the conversation at that point.The middle

In the middle of a conversation, there are really two things that you need to be able to keep in mind at any one time. The first is that you want to be listening for topics to talk about, conversational branches that are interesting to you and the other person, commonalities that you have, ways of drawing the conversation in deeper through shared topics. You can actually preload some of this work before the conversation by thinking about some topics you might have in common with that person, or preparing some topics to talk about where you feel comfortable in that space. Now, I know that this sounds quite contrived, but the research actually shows that when the conversational partners prepare topics to talk about both people in the conversation find it more rewarding, despite having predicted before the conversation that it would be awkward. In addition, the research shows that all you really need to do is think for 30 to 60 seconds about what you want to talk about. I'm not saying to prepare a list two miles long and research on Wikipedia for four days, but even giving a little bit of thought to what might be an interesting topic of conversation to talk about can go a long way.

Another powerful tool in a conversation, and this is borne out by many studies, is the asking of questions. Now whats important to note is that these are not any random questions; they're follow up questions. They are pulling on what the other person has already said, referencing prior threads in the conversation (which is also trust and relationship building because it shows that you are listening to the other person to care about what they had to say), and also follow-up questions to whats being discussed at that moment. I'm hesitant to give an exact number here because people become too mechanical about this, but it seems that a good ratio is two follow-up questions for every one statement that moves the conversation forward.

The research calls this type of conversational tactic active construction. This is when you're enthusiastic, curious, and elaborative, inviting your partner to go deeper into their experience. An example of this type of response might be: “That's amazing! How did you pull that off? What was the best part?” By showing this type of active interest in your partner, you signal to them that you care and their experience matters to you.

Another variation on this is collaborative riffing. You're not responding so much as you're building off what they said. Instead of “I went to Paris” → “I went to Rome,” riffing looks like: “I went to Paris” → “That reminds me, we should explore what makes European café culture so different from American coffee shop culture” → “Yes! I wonder if it's about pace or about the architecture...”. You can do this in four steps:Build rather than branch: Add to their idea instead of introducing your parallel exampleUse callbacks: Reference earlier shared moments (“Like we were saying about...”)Increase pace: Minimize pauses between turns to create flow momentumEmbrace vulnerability: Share unpolished thoughts to invite co-creation

I think many people will do these things automatically as they become more interested and invested in their partner, but it's important to see the tactics outright for when you find yourself flagging on things to say or talk about.The end of a conversation

The first thing to say here is that most people underestimate the length of a good conversation. The research shows that people inevitably end their conversations too early because they're afraid they're going to run out of things to talk about. They anticipate awkwardness and therefore choose to end the conversation early. If you're conducting conversations in the right way, though, then they are generative; there's not a finite amount of things to talk about. Every branch you walk down actually provides you with new material. A good rule of thumb is that you should continue the conversation for as long as you're enjoying it or until there's a reason to stop.

When you do decide to end the conversation there are a few ways to make a graceful exit:Reference shared content (“I really enjoyed hearing about your project. Let's definitely stay in touch about that”),Propose concrete next steps (“I'd love to continue this. Are you free for coffee next week?”),Acknowledge time constraints directly rather than manufacturing excuses (“I need to head out, but this was great”)What this means for you

If you dread conversations with anyone you're not close to, it's likely because you were utilising old and broken conversational tactics like mirror questions, shallow small talk, and premature exits. When you make some key changes though, like creating opportunities to go deeper, asking good questions, active construction and riffing, you suddenly realise that there were all these amazingly interesting people around the entire time!

If you remember one thing, it's that conversations should be generative, not extractive. They should create something: a shared understanding, a common reality, a connection, that wasn't there before.

Start with genuine curiosity. Ask follow-up questions. Share yourself reciprocally. Build together rather than taking turns. Listen more than you prepare to speak. And when it feels good, keep going. It's probably going better than you think.