If you sell through Instagram DMs, you have probably met this rule without being told its name. You load a tool, you find the thread of someone who asked about your program in the spring, you write the follow-up, and the send comes back refused. Nothing is broken. The software is doing what the platform tells it to do.
That rule is the standard messaging window, and almost everything written about it is either vague or wrong in one specific way: it describes the window as a limit on Instagram DMs in general, when it is a limit on one particular way of connecting to them. The difference decides whether your old conversations are reachable or not.
Here is what the window is, what opens and closes it, what each documented exception actually permits, and where your own account sits in all of this.
What is the Instagram 24-hour messaging window?
When a person acts on your professional account, a standard messaging window opens. For as long as it stays open, software connected through Meta's official business messaging can send that person messages freely. When it closes, ordinary sending stops.
Meta states the duration in its own Messenger Platform documentation, and the sentence is worth reading literally:
For most user actions, it lasts 24 hours. The window's length depends on the user action that opens it. Meta, Messenger Platform documentation
Two things follow from that wording. The clock is attached to something the other person did, never to something you did. And the duration is a default rather than a constant, which is why the exceptions further down are not loopholes but documented variants of the same mechanism.
Instagram and Messenger run on the same messaging platform, so the rule reads the same on both. What changes between them is which features Meta makes available, and sales follow-up has never been one of them.
What opens the window, and what closes it
The list of qualifying actions is published, and it is shorter than people assume. Sending you a message opens the window. So does tapping a call-to-action button inside a conversation, commenting on one of your posts, reacting to a message, or arriving through an ig.me link into an existing thread. Meta sets all of this out in the same Messenger Platform documentation alongside the 24 hours.
Every one of those entries describes the other person doing something. That is the whole design. The window is Meta's way of saying that a business may talk when it has been spoken to.
Closing is simply the clock running out. Since each new action from that person counts as a window-opening action in its own right, an active back-and-forth never hits the limit, which is exactly why the rule stays invisible until the day you try to reopen something old. A thread that went quiet in March has had no qualifying action since March, so for any tool bound by this mechanism that person is simply out of reach.
This is also why a tool's advertised sending capacity tells you very little. Capacity is about how fast a tool is allowed to send. The window is about whether it is allowed to send at all.
The three exceptions, and what each one really allows
There are documented ways to send outside the standard window. All of them exist for customer service, logistics and support, and the constraints attached make that obvious.
| What the person did | How long a tool can write | What it may send | Where Meta documents it |
|---|---|---|---|
| Sent you a DM | The standard window, 24 hours | Anything, promotional content included | Messenger Platform |
| Commented on a post | Within 7 days of the comment | One single message, and nothing more unless they reply | Private replies |
| Clicked an ad, then wrote | Up to 7 days from that first message | Anything | Messenger Platform |
| Nothing recent | No window at all | Only a tagged message, non-promotional | Messaging policy |
The human agent tag is the one people cite most often when they hope the window has a back door, and the documentation is specific about its shape:
the Human Agent tag allows a business representative to manually respond to a person's messages within a 7-day period Meta, Messenger Platform documentation
Read the two load-bearing words: a representative, responding manually. It covers a human picking up a support thread an automation could not resolve. It is not a license to schedule a promotional follow-up a week later, and Meta's messaging policy adds the ceiling that applies to every tag: messages sent outside the window must carry non-promotional content, and using a tag outside its approved cases can cost an account its ability to send.
The comment exception is tighter still, and tighter than most comment-to-DM marketing suggests. Meta's page on private replies sets three conditions at once: the reply can go out within 7 days of the comment, only one message may be sent to the person who commented, and the conversation can continue only if they answer. One message, then silence unless they choose otherwise.
The exceptions have also been narrowing rather than widening. On April 27, 2026, three tags that businesses leaned on for order and event updates stopped working and began returning an error instead, per the Messenger Platform documentation. Anyone whose follow-up plan rests on a tag is building on ground Meta moves.
Why Meta built the window this way
The rule makes more sense once you look at what it prevents. Without it, any business with messaging access could message everyone who had ever touched its account, forever, at no cost. Instagram DMs would carry the volume that killed the usefulness of email.
So Meta protects the inbox the way it protects the feed, by watching behavior. Its spam policy targets repetitive content sent in bulk, and the Community Guidelines describe conduct rather than quotas. The window is the same philosophy expressed as a timer: consent is recent or it is absent.
Worth saying plainly, because it gets muddled in both directions: this is a product rule about one route into Instagram, not a verdict on whether contacting an old lead is acceptable. Replying to somebody who once wrote to you is ordinary business conduct. What actually puts an account at risk is volume, repetition, and writing to strangers.
What the window does not cover
Here is the part that changes what you do on Monday.
The window governs software connected through Meta's official business messaging. It does not govern you. When you open Instagram on your phone and answer a conversation from eighteen months ago, no window is consulted, because you are not a business integration making a request. You are a person using an app.
That asymmetry is not a trick or a gray area. It is the plain consequence of the rule applying to a specific route, and it is why your DMs hold two populations that behave completely differently. Recent threads are reachable by essentially any tool. Everything older is reachable only by you, by hand, one scroll at a time.
Which matters, because the older group is usually both larger and warmer. Those people already know your offer, already had a conversation with you, and never said no. They said later, and later never got a reminder. That stock is the subject of its own piece on the leads still sitting in your old conversations.
What it means if you sell in your DMs
Three practical consequences, and the first one saves the most time.
When a tool refuses to send, stop debugging. A refused follow-up to a quiet lead is usually the rule working, not a fault, and no amount of reconnecting will change it.
When you evaluate a tool, ask how it connects before you compare features. Anything built on Meta's official business messaging inherits this window whatever its interface promises. That single question separates the market more cleanly than any feature list, which is the ground ManyChat and Ficiel actually differ on.
And when you plan your follow-up work, sort your inbox by reachability rather than by intent. The threads you most want to reopen are precisely the ones no automation can touch, so they need a different approach instead of a better tool of the same kind.
How Ficiel approaches it
Ficiel does not try to win an argument with Meta's rule. It stands somewhere else entirely: you connect with your own account, the way you do on your phone, so the conversation from eleven months ago is as reachable as this morning's.
What that buys is access to the part of your history nobody else can read. On Essentiel, Ficiel rereads 24 months of it, isolates the conversations that stopped without a decision, and puts them in front of you with their context.
Kord is Ficiel's AI assistant: it rereads your conversations, finds the leads who went quiet, drafts follow-ups in your own voice, and a human always approves before anything is sent.
That approval step is not a formality, and it is also what keeps the behavior ordinary. A follow-up you read before it goes out is a person answering their messages. Ficiel holds a sending pace to match, 15 follow-ups a day per channel on Essentiel, spaced rather than fired off together.
Two limits worth stating. Ficiel never writes to strangers, so everything starts with somebody who contacted you at some point. And the stock is finite: once your old conversations have been worked through, month two looks nothing like month one, which is better to know before you start than after.
FAQ
What is Ficiel?
Ficiel is an AI inbox for coaches and setters: it surfaces old Instagram, WhatsApp and LinkedIn conversations worth following up, drafts the message, and nothing is sent without your approval. Because you connect your own account rather than a business integration, the standard messaging window is not what decides who you can write to.
Does the 24-hour window apply when I reply from the Instagram app?
No. The window is a constraint on software connected through Meta's official business messaging. Replying by hand from your own account, on your phone or on a browser, is not subject to it, which is why you can answer a two-year-old thread whenever you like.
What is the human agent tag?
A documented exception that lets a business representative respond manually to a person's messages within a 7-day period. The wording matters: it covers a human answering, usually in a support escalation, and messages sent outside the standard window have to be non-promotional. It is not a way to schedule a promotional follow-up a week after the conversation stopped.
Can I message someone who commented on my post?
Yes, once. A private reply can be sent within a 7-day stretch after the comment, and only one message is allowed to the person who commented. After that the thread goes nowhere unless they answer you, at which point the standard window takes over.
Does the window reopen if the person messages me again?
Yes. Every qualifying action by that person opens a window of its own, and sending you a message is top of Meta's published list. That is why a live conversation never runs into the limit and why the rule only shows up when you reach back into an old one.
Why was my follow-up refused even though the conversation looks fine?
Most likely because the last thing that person did is older than the window, and your tool is bound by it. The thread still displays, the history is still there, but there is no open window on it. The refusal is the rule, not a bug, and the fix is not a different tool of the same kind.
The one-message limit on comment replies leaves a list behind, and what to do with the people who claim your freebie and never answer covers how to reopen those threads yourself.
