There is a specific promise that agents break more than any other, and it is not a big one. It is let me check on that and come straight back to you. Nobody makes it dishonestly. It gets broken because the next chat arrives, and the one after that, and by the time the queue clears there is nothing anywhere that remembers you said it.
The follow-up card is a timer attached to a visitor. You set it in the middle of a conversation, in about two seconds, and it comes back and finds you. This guide covers what it holds, how it gets your attention, and the two cleanup rules that decide when a reminder stops existing.
The promise nothing else tracks
Plenty of things in the console track the present. The wait indicator knows when a visitor is sitting on an unanswered message. The notes card holds durable facts about a person. The commitments card reads your outgoing messages and pulls out the promises you have made in the words you used.
None of them covers a future-dated obligation to someone you have already replied to. The moment you send give me ten minutes, the wait indicator is satisfied, because you replied. The chat looks healthy. That is precisely the gap where the promise falls through.
It is worth being clear about how this differs from the commitments card, because they sound similar. Commitments are detected — the console reads your sentences and infers a promise. Follow-ups are declared — you set them, deliberately, with a time. Detection is broader and catches things you would not have thought to log. A declared reminder is the one that actually chimes.
What a reminder carries
Setting one is a two-part gesture: pick a time, optionally type a line. The quick picks are five, fifteen and thirty minutes and one hour, with a custom field for anything else. The note is capped at two hundred and forty characters.
Alongside your note, the reminder stores who it is about — display name, email and session id — captured at the moment you create it. That snapshot matters more than it sounds. It is what lets the reminder still make sense forty minutes later when the chat has ended and the tab has closed, and it is why the reminder can still tell you who it was about when there is no live conversation left to point at.
It also means the snapshot does not update. If a visitor gives their email after you set the reminder, the reminder carries the name it had when you set it. Not a problem in practice, but it explains the occasional reminder that says Visitor rather than a name.
How it tells you
When a reminder comes due, three things happen, and only one of them is on by default.
The pill on the right rail turns red and pulses. That is the reliable one; it needs no permission and no configuration.
A chime plays — but only if you have turned sound on. It is off by default, and the toggle is a small bell in the card footer. It is a two-tone beep generated in the browser rather than an audio file, so there is nothing to load and nothing to block. If you work with the console in a background tab, turning this on is the single change that makes the feature reliable rather than decorative.
A system notification appears, but only if your browser has already granted notification permission for the console. The card deliberately never asks for that permission itself — a permission prompt triggered by setting a reminder would be a jarring interruption at exactly the wrong moment. If you have never granted it, this step silently does nothing, and the pill is doing all the work.
The practical reading: if you rely on follow-ups, spend ten seconds turning on the chime and allowing notifications. Otherwise the feature only works while you are looking at the console, which is the situation it was built to rescue you from.
Snooze, done, and the seven-day drop
From the expanded card, a due reminder gives you three moves. Jump takes you to the visitor’s tab if the chat is still open. Done marks it finished. Snooze pushes it out again.
Snooze genuinely re-arms it — the new time is set and the already-fired mark is cleared, so it will chime again properly rather than sitting there silently past due. You can snooze as many times as you like.
The cleanup rules are where a reminder stops existing, and both are silent:
- A reminder you marked Done disappears twenty-four hours later.
- A reminder you never dismissed is dropped once it is more than seven days past due.
The second rule deserves a moment. A reminder you ignored for a week is deleted on the assumption that it has been overtaken by events, and there is no record that it existed. This is a reasonable default — a console carrying a nine-month backlog of stale reminders is a console nobody reads — but it does mean the card is not an audit trail. If a follow-up genuinely matters a week later, it needed to become a ticket, not stay a chime.
There is also a ceiling of two hundred live reminders, and if you somehow pass it the oldest completed ones are discarded first.
It only fires while the console is open
Reminders are stored in your browser and checked by the console itself. There is no server-side timer, which has one consequence you should plan around: nothing fires while the console is closed.
The reminder is not lost. It is stored with its due time, and when you next open the console it is read back, noticed as overdue, and it chimes then — assuming it is still inside the seven-day window. Set a reminder for an hour, close the console, come back tomorrow, and it will greet you on arrival.
What it will not do is reach you at the moment you asked for while you are away from the console. For a reminder set during your shift, for later in the same shift, that distinction never comes up. For anything that needs to survive going home, use a real scheduled channel instead.
The card also does not re-chime for reminders it has already announced. It remembers which ones have fired, so reloading the console mid-shift does not set off every alert from the last three hours at once.
Writing a reminder you will understand later
This is the craft half, and it is where most of the value actually sits, because the mechanics above are only as good as the sentence you type into them.
The reminder arrives when you are in the middle of something else. Reading it should not require reconstructing the conversation. The note that fails is follow up. The note that works names the thing and the action: tell Anna whether the March invoice can be reissued.
Two habits make the difference. First, write the note for a stranger — assume you will have no memory of the chat, because in forty minutes and nine conversations you will not. Second, set the time to when you will actually have the answer, not when you promised to reply. If you told a customer ten minutes and you are waiting on a colleague, twenty is the honest number, and a reminder you snooze three times is one you set wrong.
The other discipline is telling the visitor the same thing you told the timer. A reminder is a private commitment; it does not communicate anything to the customer. I will come back to you within fifteen minutes plus a fifteen-minute reminder is a promise kept. A reminder with no message is just a well-organised silence, and the visitor experiences it identically to being forgotten.
How MyLiveChat fits
The follow-up card is part of the agent console and needs no setup. Ctrl+Shift+U opens it, or click the pill on the right rail. Reminders are yours alone — they live in your browser and are not visible to the rest of your team.
That personal scoping is the boundary to keep in mind. Follow-ups are the right tool for the ten-minute promise you intend to keep yourself in the next hour. They are the wrong tool for anything that has to survive your shift, be picked up by a colleague, or be answerable to later. For those, the promise belongs somewhere the whole team can see it.
Used for what it is, it removes a small, constant, entirely avoidable category of failure: the customer who was told someone would come back, and waited, and nobody did.