What do you say to start a conversation with a stranger?
Most advice treats the opener as the hard part. It is not. What separates people who have these conversations from people who don't is almost never the first line. It is whether one got sent.
The bar is lower than you are setting it
Asked whether they would strike up a conversation with a stranger, most commuters said no — in one setting 93% of them. People actually assigned to do it had better conversations than they had predicted, and none of them was given a clever opener. They were told to start a conversation and find out something about the other person. That was the whole technique.
What we think works in text
The move that is not about the opener
The strongest evidence here is about what you do after they answer. People who asked more questions were better liked, and the effect came from follow-up questions — the ones that pick up what the other person just said. That is most of the skill, and it is one move.
If nothing comes back
Here the research runs out. Every study above put two people together and started the clock. None covers the step that comes first online: being picked.
So the explanation that feels obvious from the inside — that it was something about you — is the one you have the least evidence for. Send another to somebody else. That is the only move that produces new information.
Where this comes from
Most people say they would avoid starting a conversation with a stranger, and the reasons they give are that they won't know what to say or won't be welcome. Neither survived contact with the experiment.
What the study actually did
Across the settings studied, the majority said they would avoid talking to a stranger — in one context 93% of them. Participants who were then randomly assigned to do it were given no script and no opener, just an instruction to start a conversation and find out something about the other person, and reported a better experience than those who kept to themselves.
What it doesn't show. That any particular opening line works, because none was tested. The three shapes suggested on this page are ours and are not a finding. What the research supports is narrower: that the reluctance is near-universal, that the stated reasons for it did not hold up, and that no technique was needed to get the result. We have not found these experiments run on anonymous text chat, so we are extending a finding past the setting it was measured in.
Epley, N., & Schroeder, J. (2014). Mistakenly seeking solitude. Journal of Experimental Psychology: General, 143(5), 1980–1999 ·
10.1037/a0037323
What happens after they answer matters more than the opener: people who ask more questions, and follow-up questions in particular, are better liked.
What the study actually did
Three studies of live two-person conversations — online get-to-know-you chats, and real face-to-face speed dates. Across them, question-asking predicted how much a partner liked the asker, and the effect ran specifically through follow-up questions: the ones that pick up what the other person just said.
What it doesn't show. That questions make a conversation good for you. Liking is the outcome measured, and somebody who asks everything and answers nothing can be well liked and have disclosed nothing. The settings were online text chat and in-person speed dating, both with an identity or a possible second meeting attached.
Huang, K., Yeomans, M., Brooks, A. W., Minson, J., & Gino, F. (2017). It doesn't hurt to ask: Question-asking increases liking. Journal of Personality and Social Psychology, 113(3), 430–452 ·
10.1037/pspi0000097
