Almost every awkward conversation we have with a new MConnect customer comes back to one rule. It is not a pricing rule and it is not a rate limit. It is a timer.
How the timer works
When a WhatsApp user sends your business a message, a customer service window opens and runs for 24 hours. Inside that window you can reply with free-form content — text, images, documents, quick replies — with no pre-approval of any kind.
If the user messages you again before the timer expires, it resets to a fresh 24 hours. It does not extend, and it does not accumulate. Two messages an hour apart do not buy you 48 hours; they buy you 24 hours measured from the second one.
When the window closes, free-form messaging stops. From that point you can only re-open the conversation with a pre-approved message template.
Why this is the rule that matters
Most businesses arrive with a mental model borrowed from email: compose something, press send, it arrives. WhatsApp does not work that way, and the difference is not cosmetic.
It means the shape of your messaging is decided long before you write any copy:
- Anything reactive — support, order questions, a human picking up a thread — happens inside the window and can be written on the spot.
- Anything proactive — a shipping update, a reminder, a campaign — happens outside the window and must exist as a template that was submitted and approved days earlier.
You cannot improvise the second category. If it is Friday afternoon and you want to tell 4,000 customers about a delivery delay, the template you need either already exists or it does not. Approval usually takes minutes, but it can take considerably longer if a template gets flagged for manual review — which is precisely when you are least able to wait.
The practical consequence: build your template library before you need it, not when the incident starts.
Automation is allowed, escalation is required
You may use bots and automated replies inside the window. WhatsApp’s business policy is explicit that automation is permitted — but it also requires a prompt, clear and direct path to escalate to a human.
This is worth taking seriously, and not only for compliance. A chatbot that cannot hand over is a chatbot that generates a second, angrier inbound message. In MConnect we treat escalation as a first-class part of a flow rather than a fallback: any automated branch that cannot resolve the request should route to the team inbox with the conversation history attached, so the agent who picks it up is not asking the customer to repeat themselves.
A template is not consent
The most expensive misunderstanding we see: teams assume that because a template is approved, they may send it to anyone.
They may not. Template approval and customer opt-in are separate requirements. Approval means Meta has reviewed the content. Opt-in means the recipient agreed to hear from you. You need both, and the second one is your responsibility to collect and to be able to evidence.
What changes in October 2026
Meta has announced that from 1 October 2026, service messages sent inside the 24-hour window become chargeable on a per-message basis, with rates varying by market and broadly aligned to utility and authentication rates.
If that lands as announced, the economics shift in a specific direction: the window stops being a free zone and starts being a metered one. Conversations that today cost nothing to service will carry a per-message cost, which makes resolution efficiency — fewer round trips to the same outcome — a line item rather than a nicety.
We would rather you confirm this against Meta’s own rate card for your market before you rebuild a budget around it. Announced pricing has moved before.
What to do with this
Three things, in order:
- Audit what you send proactively. Every one of those needs a template, approved in advance.
- Separate your opt-in record from your template library. They are different systems and they fail differently.
- Measure round trips per resolution, not just volume. Whatever happens with service-message pricing, a conversation that takes six messages to resolve what could have taken two is already costing you an agent’s time.
If you want to see how the template side of this constrains your writing, we drew the character budget to scale on our home page — the body is 1,024 characters, and it is smaller than it sounds.