What response time actually tells you — and what it doesn't

Response time tells you how far a reply departed from what is normal for that particular person, and it tells you nothing at all in isolation from that baseline.

This is the entire method, and it is why the advice you have already read did not help. A four-hour gap from someone who has replied within four hours for eight months is unremarkable. The same four-hour gap from someone who has never once taken more than twenty minutes is a genuine change worth noticing. No number in the abstract — not twelve hours, not a day, not “before bed” — can distinguish those two cases, because the information is not in the gap. It is in the distance between the gap and the person’s own history.

Everything else on this page follows from that. Before you can judge a delay you need a baseline, and you already have one: it is sitting in your thread, in the weeks before whatever it is that brought you here.

Scroll back to a stretch of ordinary conversation from before your concern started, and note how long replies typically took during it.

Ordinary means no argument, no crisis, no birthday, no holiday — a week that was not about anything. Two or three such weeks is plenty. You are not producing a statistic; you are answering one question: what did normal look like here? Most people can answer it in two minutes of scrolling, and most are surprised by the answer, because remembered reply times are heavily distorted by the replies that stung.

Two properties of the baseline matter more than the average, and both are easy to read off the thread directly.

A genuine outlier is a reply that falls well outside the person’s usual spread, in their usual fast window, with no visible reason — and the useful ones repeat.

All three conditions matter. Outside the spread rules out the ordinary variation every conversation has. In the fast window rules out the daily shape. No visible reason means you have already checked the thread for the obvious explanation, which is present more often than people expect: a message sent while they were mid-conversation about something else, a stated plan that evening, an argument the day before.

The repetition condition is the one most often skipped and the one that carries the most weight. A single outlier is a bad day. Outliers that cluster — always the same night of the week, always after a particular kind of question, always following contact with a particular person — are a pattern, and a pattern is something you can actually ask about, because you can describe it.

Selective responsiveness is the strongest timing signal there is, because it eliminates every explanation that depends on the person being unavailable.

Busy, asleep, driving, no signal, phone dead — every ordinary reason for a delay is a reason they could not reply to anyone. Visible activity elsewhere during the same window rules all of them out at once, which is why this observation does more work than any amount of measuring gap length. The person was there. The reply to you did not happen.

Be careful about what it establishes, because it is narrower than it feels. It shows availability, not motive. The commonest cause of a message being skipped while others are answered is not deception but difficulty: your message asked something that required a real answer, and answering the easy messages first is what most people do with a hard one. That is avoidance, and avoidance is worth raising — but it is a different conversation from the one you may be bracing for, and it is much more likely.

Response time cannot tell you why, and every use of it that matters ends at a question rather than at a conclusion.

It cannot separate a person avoiding a difficult subject from a person avoiding you. It cannot separate someone losing interest from someone in a hard month at work, and those two produce almost identical patterns — longer gaps, shorter messages, fewer questions asked back. It cannot detect a lie; the claim that deception adds a measurable delay comes from laboratory conditions with none of the interruptions a real phone has, and it does not survive contact with a working day.

What it can do is establish that something changed, and when. That is genuinely valuable, because “you’ve been different since about the start of the month” is a real observation that a person can engage with, while “you’re being distant” is one they can only deny.

Find the last stretch that matched the old baseline and the first stretch that clearly did not; the boundary between them is your date.

Dating the change is more useful than measuring its size, because a date can be checked against the world. Most pattern changes line up with something — a new job, a house move, a family illness, a change in shifts, an argument, a new person in one of your lives. Finding that the change began the week their contract started answers the question completely and costs one minute of scrolling.

Resist the urge to date it precisely. You are looking for the week, not the hour. False precision here is actively harmful: a boundary pinned to a specific evening invites you to build a theory around whatever happened that evening, and the resolution of your data does not support it.

Describe the change, not the measurement, and ask once.

“You used to text through the day and lately it’s mostly evenings — has something changed?” is a sentence the other person can recognise, confirm, correct or explain. They can tell you their shifts moved. They can tell you they have been struggling. They can tell you they hadn’t noticed. Every one of those is a real answer and every one of them is more likely than the thing you are afraid of.

Presenting numbers does the opposite. A tally of measured reply times moves the conversation onto whether the tally is fair, whether you have been counting, and how long you have been counting for — and that conversation is about you. It also teaches the other person that their replies are being scored, and the reliable effect of that is not more honesty. It is more careful management of the thing being scored.

The one-question limit is not politeness. It is what keeps the answer informative: asked once, you get their account. Asked repeatedly, you get the account they have learned produces the least friction.

Stop when you have your answer to “is this different from normal”, because that is the only question timing can answer.

If the answer is no, you have gained something real — the thing that was bothering you was not in the thread — and continuing to measure will eventually manufacture a concern, because any metric watched closely enough produces a bad day. If the answer is yes, the next step is a conversation, not more data. There is no amount of additional timing that converts a pattern into a reason.

If you find you cannot stop, that is worth taking seriously on its own terms and is covered honestly in what to do about being left on read, where the same loop shows up in its most concentrated form.

• Left on read: what it means and what to do — Being left on read establishes that a thread was opened and nothing else. What a read receipt can and cannot report, why the expensive messages are the ones that sit, whether to send a second message — and what to do when the re-reading has become the problem.
• Hours to reply, but active on Instagram — Activity during a gap eliminates busy, asleep, driving and flat battery at a stroke — and establishes availability, not motive. What an active indicator actually reports, how to rank a green dot against a posted story against a reply to someone else, and why scrolling is cheaper than answering.
• Dry texting: signal or personality? — Dry texting only carries information when it is a change. What actually makes a reply dry, who is doing the conversational work, why plenty of interested people write in fragments — and what it means if you are the one being called a dry texter.

How long should someone take to reply to a text?

There is no correct reply time, and any page that gives you a number is guessing. Reply speed varies enormously by person, by job, by time of day and by how the two of you have always texted — a four-hour gap is unremarkable from someone who has always replied in four hours and is a genuine change from someone who has always replied in four minutes. The only reply time that carries information is one that departs from that specific person's own established pattern.

Does a slow reply mean someone is losing interest?

A slow reply on its own does not mean someone is losing interest, because a single gap has too many ordinary explanations to distinguish between — work, sleep, driving, a flat battery, or simply not having decided what to say. Loss of interest shows up as a sustained shift in the pattern rather than a single slow reply: gaps that get longer week over week, replies that get shorter, and questions that stop coming back. One slow reply is noise; a trend across weeks is a signal.

What is a texting baseline and how do you find yours?

A texting baseline is the reply time that is normal for one specific pair of people, found by looking at how long replies took over several weeks before whatever prompted your concern. You establish it by scrolling back to an ordinary stretch of the conversation — one with no argument and no unusual event — and noting the typical gap between a message and its answer. Everything after that is measured against that number, not against a general rule about how fast people should text.

They replied to someone else but not to me — what does that mean?

Selective responsiveness means the person was available and did not reply to you specifically, which is a materially stronger signal than a plain delay because it removes the ordinary explanations. Being busy, asleep or out of signal cannot explain replying to one person and not another in the same window. It still does not tell you why: the message may have been one they did not know how to answer, or one they were avoiding rather than one they did not see.

Can you tell if someone is lying from how fast they reply?

No, reply speed cannot tell you whether someone is lying, and this page is not a lie detector. Claims that liars take measurably longer to answer come from controlled laboratory conditions that have almost nothing in common with a phone in a pocket during a working day, where the same delay is produced by a meeting, a commute or a difficult question asked at a bad moment. Reply timing can show that a gap is unusual for that person, which is a reason to ask a question, not evidence of deception.

Is it worth timing someone's replies?

Timing replies is worth doing once, to settle whether a specific gap was actually unusual, and it stops being worth doing the moment it becomes a habit. A single check answers a real question: is this different from normal, or does it only feel different? Ongoing measurement answers nothing further and reliably makes the person doing it feel worse, because a metric watched continuously will always produce something to worry about.

How do you bring up someone's reply times without starting a fight?

You raise the change rather than the number, and you ask about it once. Saying “you used to text back through the day and lately it's the evenings, has something changed?” describes something the other person can recognise and respond to. Presenting a tally of measured gaps moves the conversation onto whether your measurements are fair, which is a conversation about you rather than about the thing you actually wanted to ask.