Loading…
Connecting a Facebook Page fails for one reason far more than any other, and once it is connected, the window rules decide what your automation is allowed to send.
Meta grants the permission to send and receive messages to the Page, and it grants it through a person who holds a role on that Page. Connecting therefore needs the account doing the connecting to have that role — admin, or a task set that includes managing messages. Being the person who created the Page is not the same thing, and it is the mismatch behind most failed connections.
The confusing case is a Page that has been moved into a Business Manager, or one where somebody was given access to post but not to messages. Both look like a Page you can obviously use, and both fail at the messaging permission. Check the Page's roles before assuming the connection is broken.
Once a person has messaged the Page, you can reply freely — text, images, buttons, templates — for 24 hours after their most recent message. Your own messages do not extend it. Ten replies from you and none from them leaves the window exactly where their last message put it, which is the same rule WhatsApp applies and the same surprise it produces.
Outside the window, a message needs a tag that Meta recognises, and each tag has a narrow permitted use: a confirmed event, an update about a purchase that already happened, an account change. A promotion sent under any of them is a policy violation, and the platform does look. If what you want to send does not fit a tag, the honest answer is that it waits until the person writes again.
One tag extends the window to seven days: the human agent tag, which exists so that a person can follow up on something that needed looking into — an order that had to be checked, a question that took a day to answer. It is the escape hatch operators genuinely need, and it is also the one that gets misused.
It is for a human's reply. Using it to send automated messages is what gets access to it revoked, and the revocation is for the app, not the conversation. The product applies it only when a person is replying from the inbox, and the automation never sees it — which is deliberate, not a limitation.
Ice breakers are the questions Messenger offers before the person has typed anything — up to four of them, and they are the first thing a new contact sees. The persistent menu is the always-available list of actions in the conversation, there throughout rather than only at the start. Both are worth setting because both replace a blank screen with a prompt.
Each one, when tapped, sends a payload rather than text. That is what makes them automatable: match the payload in a button-tapped automation and the answer arrives instantly, every time, without a keyword having to be typed correctly. The best ice breakers are the three or four questions you already answer most.
The same two features exist on Instagram DMs, served by the same platform, which is why they are configured in the same place regardless of which channel the Page is connected to.
A Page has an inbox of its own, and Meta's handover protocol decides at any moment whether a conversation belongs to that inbox or to an app connected to the Page. Left alone, a conversation started in the Page inbox stays there, and an automation connected through an app never hears about it.
If you want everything in one place, hand conversations to the connected app — the setting exists on the Page's connection in the product — and the Page inbox becomes a fallback rather than a second place to check. Two inboxes for one Page is how a message gets answered twice, or not at all.
Almost always because the account connecting does not hold a role on the Page that includes managing messages. Meta grants the permission through that role, and having created the Page — or being able to post to it — is not the same thing. Check the Page's roles first.
24 hours from the person's last message to the Page. Your own replies do not extend it. Outside the window only a message carrying a tag Meta recognises for a narrow purpose will send, or a human agent reply within seven days.
A tag that extends the window to seven days so a person can follow up on something that needed looking into. It is for a human's reply, and using it for automated sending is what gets an app's access to it revoked.
Up to four. Each sends a payload when tapped rather than typed text, which is what lets a button-tapped automation answer it reliably. The same feature exists on Instagram DMs through the same platform.