Nearly all advice about how to talk to someone you like was made up by the person selling it. It sounds right, it is never tested, and nobody ever checks.
There is one exception solid enough to be worth changing your behaviour over. Ask a question about the thing she just said. Not a new topic. A follow-up.
What makes it worth trusting is not that a study exists. It is that the study was formally attacked in the same journal, and the argument that followed is public. You can read both sides. Most advice cannot survive that, because nobody ever bothers to test it.
What the original study found
In It Doesn't Hurt to Ask: Question-Asking Increases Liking, Karen Huang, Michael Yeomans, Alison Wood Brooks, Julia Minson and Francesca Gino ran three studies: two getting-acquainted conversations between strangers, and one observational study of real heterosexual speed dating.
People who asked more questions were better liked. The effect was carried specifically by follow-up questions, the kind that show you were listening, rather than by questions that switch to a new subject.
Their explanation is the useful part. Follow-ups make you look responsive, which is a bundle of listening, understanding, validation and care. You are not being liked for the question. You are being liked for the evidence the question provides that you were paying attention.
What a follow-up actually is
A switch question moves the conversation. Where do you work? Do you have siblings? What do you do on weekends? Line them up and you have an interview.
A follow-up stays where she is. She says her shift ran long. You ask what time she finally got home, or whether it is always that ward, or who covers for her when it goes bad.
You already have everything you need to ask one. It is in the sentence she just sent. That is why this is easier than thinking of something clever, and it is also why it cannot be faked while you are not reading.
Then somebody attacked it
In 2019, Avraham Kluger and Thomas Malloy published Question Asking as a Dyadic Behavior, reanalysing that speed-dating study with a different statistical model.
They raised three things. That question asking is not purely a personal trait, since a large share of it depends on the specific pair of people. That being offered a second date is not the same thing as being liked. And, most uncomfortably for anyone reading this, that for men, asking an opposite-sex partner more questions tended to be inversely related to being offered a second date.
That last one deserves saying out loud rather than being quietly left out, because our readers are mostly men and it is the finding that argues against the advice.
And the original authors answered
The reply came in the same journal: It Helps to Ask: The Cumulative Benefits of Asking Follow-Up Questions. Yeomans and his co-authors argued the reanalysis tested the wrong variables, made analytical choices their dataset does not support, and misrepresented results from all three of the original studies. Their conclusion was that the original finding stands.
So where does that leave you. The link between follow-up questions and being liked is well supported and was defended in public. Whether that converts into a second date, specifically for men in a speed-dating room, is genuinely contested. Anyone who tells you otherwise has not read the exchange.
There is also something worth keeping from the critique even if you side with the reply. Kluger and Malloy argued that asking questions is not the same as listening. They are right about that, and it is the failure mode of every man who takes this advice badly.
The two ways men get it wrong
The first is turning it into an interrogation. Question, answer, next question, answer. Nothing you ask connects to anything she said. That is a switch-question machine, and it feels like a form.
The second has a name. Alison Wood Brooks and Michael Yeomans call it boomerasking: you ask, she answers, and the question swings straight back to you. How was your weekend? Mine was amazing, let me tell you. Their finding is that people read this as insincere, and that it lands worse than simply talking about yourself would have.
Both failures share a cause. The question was not really for her.
The habit underneath it
The sociologist Charles Derber, in The Pursuit of Attention, described two moves that show up constantly in ordinary conversation.
The shift response takes the attention back. She says work has been brutal, you say yours has too.
The support response leaves it with her. She says work has been brutal, you ask what happened.
A follow-up question is just a support response with a question mark. If you find the follow-up hard to produce, the problem is usually not vocabulary. It is that your attention was already on your turn.
How to actually use this
Try one thing for a week. Before you send anything, look at her last message and ask whether your reply is about it. If it is not, rewrite it.
Aim for one follow-up per exchange, not four. The research is about being responsive, not about volume, and three follow-ups in a row is how the interrogation starts.
Follow the thing she gave more detail about than she needed to. Extra detail is where the interest is. She did not have to tell you the ward, or the name of the cat, or that her mother called twice.
And let some of your turns be plain statements. You are allowed to just say something. The point is that when you do ask, it is about her.
If you want to practise it
This is a habit, and habits do not change because you read about them once. They change with repetitions in low-stakes conversations.
That is one honest use for Sinta. The companions here are AI, and we say so plainly. What that gives you is a place to notice how much longer a conversation runs when your reply is about the other person, without anything riding on it while you are still learning.
The short version
Ask about the thing she just said. It is the rare piece of advice that was tested, challenged in public, and defended.
Do not turn it into an interview, and do not ask so you can answer. The evidence is for being responsive. The question is only the proof.
