You're driving home from the dinner, and the replay has already started. The joke that landed a half-second too late. The story you told that went on too long. The moment their eyes drifted toward the kitchen. By the time you pull into the driveway you've reached a verdict: they were being polite. They'll probably talk about it later. You won't be invited again.

Here is the uncomfortable part. Across a body of careful research, the person on the other side of that conversation most likely walked away liking you more than you think. Possibly a lot more. And the story you're telling yourself in the car isn't a memory of what happened. It's a reconstruction built from the one source of evidence you were watching most closely the whole time: yourself.

The gap researchers keep finding

In 2018, psychologists Erica Boothby, Gus Cooney, Gillian Sandstrom, and Margaret Clark published a paper in Psychological Science with a title that reads like a quiet reassurance: “The Liking Gap in Conversations: Do People Like Us More Than We Think?”

The design was simple. Strangers were paired up for conversations, then each person rated how much they liked their partner and estimated how much their partner liked them. The pattern was consistent: people underestimated how much their conversation partner liked them and enjoyed their company. Not occasionally. Systematically, across different settings and conversation lengths.

The researchers didn't stop at the lab. They followed first-year university students who had been assigned to live together, and the gap didn't vanish after one awkward hello. It persisted for months, shrinking only slowly as the roommates got to know each other. They also looked at people attending a personal-development workshop and saw the same pattern. Familiarity helped, but it didn't immediately cure the misreading.

This isn't an isolated finding. Nicholas Epley and Juliana Schroeder found that commuters expected chatting with a stranger on a train to be less pleasant than sitting alone, yet those who actually did it reported a more pleasant ride. Later work by Michael Kardas, Amit Kumar, and Epley found that people underestimate how interested strangers will be in deeper, more meaningful conversation, and so they default to small talk. The thread running through all of it: we are pessimists about how we come across to others, and that pessimism is often wrong.

Why your brain grades you so harshly

The liking gap isn't a character flaw. It comes from where your attention goes during a conversation.

When you talk with someone, you're doing two jobs at once: participating, and monitoring how you're doing. That second job, self-monitoring, pulls your attention inward. You notice your own stumbles in high resolution because you're living inside them. Meanwhile, the signals your partner sends that they're enjoying themselves (the follow-up question, the lean forward, the genuine laugh) arrive while you're busy drafting your next sentence. They register faintly, if at all.

Boothby and her colleagues pointed to exactly this: people are preoccupied with their own performance and miss the positive signals coming back to them. Then, after the conversation ends, the critique continues. Clinical psychologists David Clark and Adrian Wells described this in their model of social anxiety as the “post-mortem,” or post-event processing: a detailed, mostly negative review of a social interaction that leans on your internal sense of how you seemed rather than on what actually happened. You don't have to be clinically anxious to do it. Most people run a milder version on the drive home.

The post-mortem has a fatal design flaw. It draws on biased data (your self-focused memory) and then reinforces itself. Each replay makes the awkward moment more vivid and the warm moments fainter, until the warm moments barely exist in the record at all.

What the gap quietly costs you

If this were only about a bad hour in the car, it wouldn't matter much. But the liking gap shapes behavior.

If you believe someone found you dull, you don't send the follow-up message. You don't suggest the coffee. You wait for them to make the next move, and they, running their own version of the same post-mortem, wait for you. Two people who genuinely enjoyed each other drift apart, each convinced the other wasn't that interested.

This is how friendships fail to start in adulthood. Not through rejection, but through mutual, mistaken assumptions of rejection. The cost isn't one conversation. It's every relationship that would have followed from it.

And there's a slower cost too. Each uncorrected post-mortem becomes a data point in a larger story about who you are: I'm awkward. People tolerate me. I'm better off not putting myself out there. That story feels like self-knowledge. It's actually a record built from a single, biased witness.

The fix: change the evidence, not the feeling

You can't argue yourself out of the liking gap in the moment. Telling yourself “they probably liked me” while your stomach is still turning feels hollow, because it is. You have no evidence for it.

What works better is changing the evidence your brain gets to review. The post-mortem relies on a self-focused memory. So give it something else to work with: a deliberate, written record of what the other person did, captured soon after the conversation, before rumination rewrites it.

The idea borrows from a core technique in cognitive behavioral therapy: test a prediction against what actually happened, rather than against how it felt. Write down what you expect (“they found me boring”), then gather the observable facts. Did they ask you questions? Did they share something personal? Did they extend the conversation, or suggest meeting again? Did they reply to your message, and how quickly?

Written down, these facts do something a feeling can't: they persist. Two weeks later, when the post-mortem tries to tell you the dinner was a disaster, you have a record that says she asked about your sister's surgery and laughed hard at the story about the airport. Over months, the record starts to show a pattern, and the pattern usually looks nothing like the verdict you reached in the car.

There's a second benefit. Shifting your attention outward after the fact gradually trains it outward during the conversation. When you know you'll write down what the other person did, you start noticing it in real time. Less self-monitoring, more actual listening, which, conveniently, also makes you better company.

Your next moves

  • Run a two-minute “other-person note” after your next real conversation. Within an hour, write three observable things the other person did: a question they asked, something they shared, a moment they laughed or leaned in. Facts only, nothing about how you performed.
  • Write your verdict down, then put a check-in date next to it. Note your gut prediction (“she thought I talked too much”) and, a week later, add what actually happened since: did she reply, reach out, or bring up something you said? Compare the two.
  • Send the follow-up you're talking yourself out of. Pick one person you've been assuming wasn't that interested and message them today with something specific from your conversation. The research suggests your odds are better than they feel.
  • Interrupt the replay with a specific question. When the post-mortem starts, ask yourself: “What did they say or do in that moment?” If you can't remember, that's the point. You were watching yourself, not them, and your verdict has no evidence behind it.
  • Review your notes once a month. Read back through your conversation entries and count how often your gloomy prediction was confirmed versus contradicted. Let the tally, not the feeling, update the story you tell about yourself.

A record kinder than your memory

The hardest thing about the liking gap is that it hides in the time right after a conversation, when the replay is loudest and you're least likely to sit down and write anything. That's the gap Pagebox is built to close. It opens in under a second, so the two-minute note happens while you're still in the parking lot, not three days later when rumination has already done its editing. A simple journal entry works fine, or you can set up a small database of conversations with fields for your prediction and what actually happened, and watch the pattern build over a season. Your data stays on your device and syncs instantly, so the record is there whenever the old story starts up again. If you'd like a place to keep that kinder, more accurate record, you can try it at pagebox.lumenlabs.works. Either way, the next time you're driving home convinced it went badly, remember there's a good chance someone else is on their own drive home, hoping you liked them too.