In a service business most enquiries arrive on WhatsApp. How to build a queue, an automatic first reply and a handoff to a person, inside Meta's 24-hour window.
Key takeaways
- The problem is not reply speed but the absence of a queue - a read enquiry is no longer marked.
- An automatic first reply buys time, provided it says when a real answer will come.
- Meta defines a 24-hour service window; outside it, approved template messages are required.
- Handoff to a person must be explicit - a bot that cannot hand off produces an angry customer.
- What decides the outcome is that every conversation ends in a status, not in a read receipt.
In Israeli service businesses most enquiries arrive on WhatsApp, and usually into the same app where the family conversation is running. The result is that enquiries get read and forgotten, there is no way to know how many there were, and nobody can say how long it takes to reply. That is an operational problem you can solve without changing the channel customers like.
What is actually broken
- No queue - a message that was read looks like a message that was handled.
- No measurement - how many enquiries there were, and how many answered within the hour.
- No history - a customer returning after a month starts from zero.
- No handoff - whoever replied is stuck with the enquiry.
- No separation between a new enquiry, an existing customer, and a supplier.
The first absence is the root of the rest. The moment every conversation has a status - new, in progress, waiting on customer, closed - all the other problems get a frame to be solved in.
The 24-hour window: the rule that shapes the design
On Meta's WhatsApp Business Platform, once a customer sends a message a service window opens in which you can reply with free-form messages. Outside that window, a business-initiated message must use a pre-approved template. This is a platform rule, and it directly affects how the process is designed.
What follows practically: a fast first reply is not only good service - it is also what keeps the conversation inside the window. And any proactive reminder or update sent the next day has to be planned as a pre-approved template, not as a message somebody writes in the moment. The full explanation of templates and the approval process is in the WhatsApp template approval guide.
The first reply: what it has to say
| What is included | Why |
|---|---|
| Confirmation the message arrived | Prevents a repeat send |
| When a real answer will come | Expectation is what reassures, not speed |
| What to send in the meantime | Saves a round of questions |
| A way to escalate a genuine emergency | So not everyone writes "urgent" |
The second row is the substance. "We will get back to you shortly" is not information and therefore does not reassure; "we will get back to you today by 5pm" is information, and as long as you meet it, repeat messages drop substantially.
Response hours and the expectation they create
Customers treat WhatsApp as an immediate channel, so an absence of reply reads differently there than in email. A small business cannot always reply immediately, and that is fine - what is not fine is failing to say what you can do.
Two things close most of the gap. First: stated response hours, appearing both in the automatic reply and in the business profile. Second: a different message inside hours and outside them, so a customer writing at ten at night gets a reply that acknowledges it explicitly.
What does not help is promising 24/7 availability you do not have. That promise creates an expectation where every deviation reads as being ignored, and in a small business it is also what has people replying on a Saturday evening out of guilt rather than out of a decision.
What to automate and what to leave to a person
Automate: acknowledgement, collecting basic details, sending recurring information (hours, address, price list), appointment reminders, and confirmation requests. These are fixed answers that do not change between customers.
Leave to a person: anything involving a non-standard price, a complaint, or a question whose wording is unclear. The simple rule - if the answer depends on judgement, a person answers. That is the same logic as in automating the wrong things, and it applies doubly here because a customer feels immediately when they are talking to a machine about something requiring understanding.
Handoff: the point where this breaks
The common failure is not a bot giving a wrong answer but a bot that cannot stop. A customer who writes three times something the system does not understand and keeps receiving a menu is a customer who has already concluded something about the business.
Two simple rules prevent it: after two consecutive misunderstandings, hand off to a person without asking. And at any point, one agreed word ("agent") hands off immediately. Equally important: when handing off, say so explicitly and state when a reply will come, or the silence reads as being ignored.
What about a message that is not a new enquiry?
A large share of WhatsApp traffic is not a lead: existing customers asking about something open, suppliers, and wrong numbers. When all of them enter the same queue, the queue stops reflecting real load and nobody trusts the numbers coming out of it.
The fix is simple and needs no system: a first-pass classification into three categories at the moment of arrival - new enquiry, existing customer, other. That takes a second for whoever is reading, and it is what lets you say at month end how many leads there genuinely were.
The second category earns something extra: an existing customer asking a question is usually asking a question that has been answered before. After a month of classification you can see which questions repeat, and that list becomes your ready answers - the largest saving available in this channel.
What gets recorded per conversation
At least five fields: who, when it arrived, what they asked, who is handling it, and the status. That is the minimum that turns WhatsApp from a channel into a work list, and it is also what lets you answer how many enquiries there were this month.
Beyond that it is worth recording the source - whether it came from an ad, the website or a referral - and that is captured most simply at the moment of arrival. How to do it without relying on analytics is in lead source tracking without analytics.
Sources
Frequently asked questions
Can this be done in the regular app?
Partly. WhatsApp Business gives quick replies and labels, which improves order. What it does not give is genuine multi-user access, history linked to a customer, and measurement - the difference is detailed in [WhatsApp Business API versus the apps](/blog/whatsapp-business-api-vs-apps).
How fast do you have to reply?
Fast enough to stay inside the 24-hour window, and in a way that meets what the first reply promised. What damages trust is not a long time but a promise that was not kept.
What about enquiries outside business hours?
Reply automatically with the time you will answer, not with "we are closed". The second message ends the conversation; the first keeps it, and it is also what keeps the window open the following morning.
Do you need an AI bot?
Not at the first stage. A first reply, a queue and a status solve most of the problem with no language model. A bot pays off when there is high volume of repeating questions with stable answers, and even then the handoff is the part that decides.
Keep reading
Related service
WhatsApp Cloud API
Templates, a two-way inbox and reminders on the official Meta API.
About the author
Yehonatan Saadia
Freelance automation, web & MVP developer
I'm Yehonatan Saadia, a senior developer who builds business automation, custom websites, and MVPs for small and mid-sized companies across the US, Europe, and Israel. These guides come from real client work, not theory.
Work with meHave a project like this?
Tell me what you're trying to automate or build and I'll tell you the fastest reliable way to ship it.