A WhatsApp shared inbox lets several staff answer one business number while every chat has exactly one owner. Without it, two people reply to the same guest, or nobody does. The fix is not more phones. It is an assignment rule: who gets each new chat, how many they can hold, and what happens when they are full.
We run guest support for more than 100 homestay units in Johor Bahru from one guest line, so this is the part of messaging we think about most. Below is what Google's first page tells you, what the WhatsApp Business app can and cannot do, and the exact rules our own inbox, AntlerChat, uses to split chats between staff.
What does Google say about WhatsApp for multiple users?
Search "whatsapp multiple users one number" and the first page is almost all software vendors (ChatDaddy, Wati, Cooby, Keaz, TimelinesAI and others). They agree on the basics: the free WhatsApp Business app is built around one phone, linked devices are a workaround, and a team that grows past a few people moves to the WhatsApp Business Platform (the Cloud API) with an inbox on top.
What they mostly skip is the boring question that decides whether a team inbox works: the rule that decides who owns a chat. That is what this article is about.
Can several people use the WhatsApp Business app on one number?
Yes, up to a point. Most guides put the free Business app at one primary phone plus four linked devices, with more devices for some paid Meta Verified accounts. Check WhatsApp's own help page for the current limit, because it has changed before.
Linked devices work well for two or three people sitting in one office. They break in three ways as the team grows:
| Problem | What it looks like with linked devices |
|---|---|
| No owner | Every device sees every chat. Two staff answer the same guest, with different answers |
| No record | The guest sees one name. You cannot tell later which colleague sent what |
| No limit | Nothing stops one person drowning while another has nothing to do |
If you are one or two people, stop here: linked devices are probably enough, and you do not need to pay for an inbox.
What changes when you move to the Cloud API?
On the Cloud API, the number lives on Meta's servers instead of a phone, and staff log in to an inbox instead of the app. Since 1 July 2025 Meta charges per delivered template message, by the country of the guest's number. Free-form replies inside the 24-hour customer service window, which opens each time the guest writes, are free, and so are utility templates sent inside that window. We are not quoting Meta's per-message rates here; check Meta's pricing page for the current Malaysian figure.
For a homestay that means: answering guests who message you costs nothing in Meta fees, while reminders you send first, outside the 24 hours, are the part you pay for.
How does our inbox decide who gets a new chat?
AntlerChat has two modes, set once for the company.
| Mode | Method | Who gets the new chat |
|---|---|---|
| Manual | Supervisor | Nobody until a supervisor assigns it to a person or a team |
| Manual | Agent claim | Nobody until a staff member presses claim |
| Auto | Fair (default for auto) | The active staff member with the fewest open chats |
| Auto | Round robin | The next person in turn |
| Auto | Random | Any eligible person |
Three rules sit under every auto method:
- Each person has a cap on open chats. The company default is 8, and it can be set from 1 to 99, or to no limit. A person can set their own cap; a cap of 0 means "never auto-assign me".
- Staff switched off on the users page are never auto-assigned.
- If everybody is at their cap, the chat is left unassigned rather than forced onto someone.
In agent-claim mode the same cap applies: a person who is full cannot claim another chat, and nobody can claim a chat that already belongs to a colleague. A chat can also be assigned to a team (a position) instead of one person, and everyone in that team sees it under their assigned filter.
What does the guest see when a person takes over?
Our AI answers first on most guest chats (we wrote about what it answers and what it hands over). When a colleague sends their first message, the guest gets one line before it: a human colleague has joined, with the colleague's first name from their staff account. It is sent once per takeover, not on every message.
From that reply on, the bot is quiet in that chat. What happens next is a company setting:
| Setting | Behaviour |
|---|---|
| Auto-resume off (default) | The chat stays with the person until someone presses "back to bot" |
| Auto-resume on | The AI picks the chat back up a set time after the colleague's last reply. The default is 30 minutes, adjustable from 1 minute to 24 hours |
A colleague who presses the "take over" button, instead of just replying, keeps the chat until they hand it back, even with auto-resume on. That is on purpose: pressing the button means "this one is mine".
Which setup fits which team size?
There is no single right answer, so here is how we would choose.
| Team | Our suggestion |
|---|---|
| 1–2 people, one office | WhatsApp Business app with linked devices |
| 3–5 people, mixed shifts | Shared inbox, manual with agent claim, cap around 5–8 |
| 5+ people or 24-hour cover | Shared inbox, auto with fair assignment, plus a supervisor watching unassigned chats |
Fair assignment counts open chats, not how hard they are. One guest with a flooded bathroom counts the same as one guest asking for the Wi-Fi password. A supervisor still has to move chats around on busy days.
What still goes wrong with a shared inbox?
Being honest about our own setup:
- Auto-assignment happens when a chat is first created. A guest who writes again later lands back in the same conversation and owner. If that colleague has left, you reassign their chats by hand.
- Unassigned chats need a watcher. When everyone is at their cap, new chats wait. If nobody looks at the unassigned list, the guest waits too, unless the AI is answering in the meantime.
- Auto-resume can cut in. With it on and a short timer, a colleague who steps away mid-conversation may come back to find the bot answering again.
- A WhatsApp Web line is not the Cloud API. Our plans include one WhatsApp Web number, which runs as a linked device of an ordinary WhatsApp number. WhatsApp can restrict numbers it sees as spam or bulk messaging, whatever tool is used. If losing a line would hurt your business, put it on the Cloud API.
What should you ask before you pay for any team inbox?
- When two staff open the same chat, what stops both replying?
- Can I cap how many chats one person holds, and what happens when everyone is full?
- Does the guest see which person answered?
- When a person replies, does the bot stop, and when does it come back?
- Is my number on the official Cloud API or a linked device?
- What do I pay Meta on top of the software?
Our own answers are the tables above. AntlerChat costs RM 450 a month, RM 4,860 for one year or RM 8,640 for two years, with one WhatsApp Web number included and RM 200 a month for each extra one, all before SST. It runs our guest support and links each chat to the AntlerHub booking. See how it works at /chat, and for the messages it sends on its own, read the 9 moments we message guests and WhatsApp check-in automation that never invents a link.