Reply speed means almost nothing alone. Timing shifts after emotion are where the real signal lives.
Five minutes or five hours. Neither means much without the person's usual pace. Compare replies across ordinary days, difficult conversations and attempts to reconnect. Calls, work and life outside the chat still matter.
a slow reply is only meaningful relative to the relationship baseline
Context-first
delay changes after conflict are often the strongest timing signal
After tension
the trend matters more than one late night or one busy day
Pattern > moment
Baseline matters
Some pairs naturally reply slowly. That is not a red flag by itself. The signal starts when the timing pattern changes in emotionally specific situations.
Avoidance often shows up as timing before words
A thread can look verbally polite while still showing a timing collapse after hard conversations, which is why reply patterns are so useful.
This metric helps explain drift
If warmth stays similar but delay spikes after friction, the relationship may be struggling with tolerance for repair rather than with affection alone.
What the response-time metric actually measures
Grey Mirror looks at average reply timing, consistency, and asymmetry between the two people. It is strongest when combined with windows in the timeline, because the story is usually about what changed rather than what the absolute number is.
Average reply speed
Consistency versus erratic timing
Asymmetry between the two sides
Timing shifts after emotionally loaded exchanges
When slow replies are not the point
Work schedules, sleep patterns, parenting load, and ordinary life constraints can all create slow replies. That is why blunt advice like “they took six hours to answer, so they do not care” is usually low quality.
The metric becomes useful when the delay is selective: fast in easy moments, slow only after tension, vulnerability, or accountability asks.
What reply-delay drift looks like
A common pattern is this: early in the relationship, both people respond quickly and stay emotionally available even after minor misunderstandings. Later, one person starts going quiet after hard moments, and repair windows stretch longer or never reopen properly.
That kind of drift is often more important than the occasional missed text. It changes how safe it feels to bring up real issues in the thread.
How to read timing with other metrics
Response time should usually be read beside repair attempts, pursue/withdraw index, affection/conflict ratio, and communication shifts. Timing tells you where the pressure lands; the other metrics help explain why.
A fast reply and a meaningful reply are not the same thing
Speed is easy to measure, which is exactly why it gets over-read. Someone can answer in ninety seconds with a single word that closes the subject, or take four hours and write the thing that actually moves the conversation forward. Only one of those is engagement, and it is not always the fast one.
Grey Mirror reads reply speed next to what the reply contained: whether it picked up the thread, asked anything back, or ended the exchange. A thread where replies got faster and shorter at the same time reads very differently from one where they simply got faster.
Reply latency measured against that pair's own baseline, not a universal number
Whether the reply continued the topic or closed it
Whether questions asked were answered or passed over
Length and substance alongside speed
Emotional messages and logistics messages deserve different clocks
Most threads carry two kinds of traffic: working out who is picking up dinner, and saying something that costs the sender to say. Averaging them together hides the thing people actually want to know.
Ten minutes to confirm a time is unremarkable. Ten hours to answer a message where someone made themselves vulnerable is a different event, even when both come from the same person in the same week. Grey Mirror separates these so a busy logistics stretch does not disguise a pattern of going quiet whenever a conversation turns serious — and so a slow logistics reply is not mistaken for withdrawal.
Who starts the conversation, and whether that changed
Timing describes what happens after a message arrives. Initiation describes whether it arrives at all. Both belong in the same read, because a thread can keep a healthy reply speed while one person quietly becomes the only one starting anything.
The measure that tends to matter is not the raw split — plenty of steady relationships have one person who simply texts first more often — but whether the split moved, and when. A balance that held for a year and then shifted inside a fortnight is a change worth looking at. A balance that has always leaned one way is usually just how those two people talk.
Share of conversations opened by each person, over time rather than in total
Whether a shift in initiation lines up with a shift in reply timing
Unanswered opening messages, which are easy to miss when scrolling
Whether the thread goes dormant unless one specific person restarts it
Silence after conflict, and the double-text question
The most common reason people go looking for a reply-time analyzer is a gap they are staring at right now, and the question underneath it is usually whether to send another message. That is a decision the thread can inform but not make.
What the record can show is whether this gap resembles previous gaps. If every past disagreement was followed by a similar quiet stretch that closed on its own, this one has precedent. If gaps after conflict have been getting longer, or if repair attempts that used to be answered now go unanswered, that is a change — and a change is a fact about the thread, not a verdict about the person.
Grey Mirror will not tell you whether to double-text, and it does not score anyone's intentions. It shows you what happened last time, how often, and how certain that reading is.
How to read the timing chart
The timing view plots each reply against the conversation's own baseline rather than against a fixed scale, so a naturally slow-moving thread does not read as a problem and a fast one does not read as healthy by default.
Two shapes carry most of the meaning. A step is an abrupt move to a new normal that then holds, which usually points at a specific moment worth locating in the thread. A slope is a gradual stretch over months, which rarely has a single cause and is easy to miss while living through it. Isolated spikes — a bad week, a holiday, a broken phone — are common and are labelled as low-confidence rather than presented as findings.
What this cannot tell you, and what happens to your thread
Reply timing records behaviour, not motive. It cannot tell you whether someone was busy, hurt, distracted, or losing interest, and it will not diagnose anyone, prove intent, or predict whether a relationship ends. Where evidence is thin, findings are labelled uncertain instead of being rounded up into a story.
The export itself also has limits worth knowing: platforms differ in whether they preserve edits, reactions, deletions, and precise timestamps, and anything missing from the file is missing from the analysis. Those gaps are stated in the report rather than filled in with assumptions.
Your conversation is analyzed to produce your report. It is not sold and not used to train public models, and you can delete an upload and its report from your account. On a Mac the app reads the thread already on your machine, so there is no export file to leave sitting in Downloads.
Frequently Asked Questions
How do you analyze reply time patterns in texts?
Reply time patterns in text messages are one of the clearest signals in a thread, because they are behavioural rather than interpretive — nobody performs a response delay for an audience. Grey Mirror measures every gap across the whole exported history, separates the two people, and controls for time of day and day of week so a night-shift schedule does not read as withdrawal. What matters is not any single slow reply but the distribution: whether one person consistently waits longer, whether the gap widened at an identifiable point, and whether slowness clusters around specific subjects. Each finding links to the messages that produced it, so you can check the pattern rather than take it on faith.
Does a slow response always mean avoidant attachment?
No. A slow response by itself is too weak. The better question is whether the slowness becomes patterned, emotionally selective, and paired with weaker repair or rising withdrawal.
What timing pattern is most concerning?
Usually a stable pattern of slower replies after conflict, reassurance bids, or accountability asks. That often signals that hard moments are no longer recoverable in the old way.
Do slower replies mean they are losing interest?
Not on their own. Slower replies are only meaningful against that thread's own baseline, and a single slow stretch is usually just life. What is worth reading is whether the change persisted, whether it appeared specifically around emotional messages, and whether initiation moved at the same time. Grey Mirror separates a repeated pattern from an isolated moment and labels how certain each reading is. It does not infer anyone's feelings.
Should I double-text if they have not replied?
Grey Mirror does not answer that, and any tool that does is guessing. What it can show is whether this silence resembles previous silences in the same thread: how long past gaps ran, whether they closed on their own, and whether gaps after conflict have been getting longer. That is context for your decision, not the decision.
Does it count a two-word reply the same as a real one?
No. Speed is read alongside what the reply did — whether it continued the topic, answered a question, or closed the exchange. A thread where replies became faster and shorter at the same time reads very differently from one where they simply became faster.
What happens to my conversation after the analysis?
It is used to produce your report. It is not sold and not used to train public models, and you can delete an upload and its report from your account. On a Mac, the app reads the thread already on your machine, so there is no export file to manage at all.
Use timestamps and ordered messages; an undated screenshot cannot supply a reliable reply interval.
02
Comparable windows
Compare the same person and similar periods rather than imposing a universal response deadline.
03
Context and return
Read availability, unanswered questions and later engagement alongside the delay.
Conceptual reading guide. This diagram explains what to inspect; it is not a customer report, diagnostic test or measured relationship score.Use this guide on the question you came with
Longer replies can show a timing change without showing reduced care. Missing timestamps or cross-platform contact limit the interpretation.