Migrate the number clients already message, or run it alongside the app. What each route costs you in chat history, and the checklist to do it in one evening.
Most brokerages do not start on the WhatsApp Business Platform. They start with the free WhatsApp Business app on a handset in the office, it works, and three years later that number is on the email signatures, the certificates, the van livery and in eight hundred clients' phones. Then the firm outgrows it: one handset, no audit trail, no way for a second account handler to answer, and nothing that survives the day the handler resigns.
At that point the question is not whether to move to the API but how to move that specific number without breaking the relationships attached to it. Meta offers two routes, and they have very different consequences. If you are still deciding whether to move at all, the app versus API comparison covers that; this is what happens on the day you commit.
The traditional path. Meta's documentation on migrating an existing number to a business account is blunt about it: you delete the account from the WhatsApp Business app (Settings, then Account, then Delete my account), wait for the number to become available, then register it with the Cloud API. Meta tells you to back up your chats first, because after deletion the messaging history cannot be recovered — and once the number is registered on the Cloud API you cannot use it in the app again unless you deregister it.
For a brokerage that is a real loss, and not only a sentimental one. The chat threads on that handset are the record of what was said about claims, adjustments and cover — the thing a compliance officer or an ombudsman will ask about years later. Deleting them and starting a clean API history is a decision, not an accident, and it belongs in writing somewhere.
Meta's newer path lets a number stay on the WhatsApp Business app and be connected to the Cloud API at the same time, through Embedded Signup onboarding for business app users. According to Meta's coexistence documentation, the business can synchronise up to 180 days of message history and its WhatsApp contacts if it consents to sharing, and from then on messages sent and received are mirrored between the app and the API.
The trade-offs are documented too. Numbers in use with both the app and the Cloud API run at a fixed throughput of 20 messages per second. Group chats, disappearing messages, view-once messages, live location, broadcast lists and the app's business tools such as catalogues and orders are not supported through the Cloud API side. Companion apps on Windows and WearOS stop working, and other linked devices have to be re-linked after onboarding. The app also has to be on a recent version, and the synchronisation has to complete within 24 hours of onboarding.
| Question | Delete and register | Coexistence |
|---|---|---|
| Existing chat history | Lost on deletion — back it up first | Up to 180 days can be synchronised, with consent |
| Keep using the phone app | No, unless you deregister from the API | Yes, both run in parallel |
| Throughput | Standard API throughput | Fixed at 20 messages per second |
| App-only features | Gone with the app | Groups, broadcast lists, catalogues unsupported via the API |
| Availability | Always available | Depends on your provider supporting the onboarding flow |
Three surprises come up repeatedly, none of them technical failures. The first is broadcast lists: a lot of small brokerages run their renewal chasing from broadcast lists in the app, and that habit does not survive the move — outbound at scale on the API goes through approved templates instead. The second is the handset itself, which stops being the place where work happens; if two people were quietly sharing that phone, the move ends that and the change has to be announced internally. The third is display name and business verification, which Meta reviews on its own timetable, so it is not something to leave until the morning of go-live.
Outbound messaging also changes shape. Once a conversation has gone quiet, you cannot simply type at the client: outside the 24-hour customer service window you send an approved template, and its category determines what you may say in it. If your team has never submitted one, read the guide to utility versus marketing template approval before you migrate, not after — an empty template library on day one is the most common reason a migration feels like a downgrade.
Before any of this, settle where the messages will live and who can get them out. Moving to the Platform means client conversations pass through a provider's systems as well as Meta's, and your UK GDPR or GDPR documentation has to reflect that: what the provider is doing on your behalf, where data sits, how long messages are retained, and how you would answer a subject access request or an ombudsman file request. A migration is the natural moment to fix this, because you are already deciding what happens to history. The companion piece on archiving WhatsApp under SYSC and ICOBS sets out what a defensible retention position looks like.
ORIS connects a brokerage number through Meta's Embedded Signup and registers it against your own WhatsApp Business Account, so the number and the client relationships stay yours rather than sitting inside an agency's account. Conversations then land in a shared inbox against the customer record, templates sync from Meta with their approval status visible, and the interface shows whether the 24-hour window on a thread is open before someone tries to reply freely. Whether the coexistence route is available to you depends on the onboarding flow your provider supports — ask that question explicitly during the demo rather than assuming it, because it decides whether your history comes with you.
The number stays the same, so their thread stays where it was. What changes on your side is the display name shown on the profile and, if you took the deletion route, the absence of the old history. Clients rarely notice the mechanics; they notice if replies suddenly stop or come from a different number, which is the argument for moving the existing number rather than starting a new one.
Yes, but on the deletion route it means deregistering the number from the Cloud API first, and you would not carry the API history back into the app. Treat the move as a one-way decision for planning purposes, and pilot with a second number if the firm is genuinely unsure.
Opt-in follows the client relationship, not the technical connection, so a client who asked you to message them has not withdrawn anything. What the move does change is scale: sending templated outbound to hundreds of contacts is a different exercise from replying to inbound chat, and marketing content in particular needs a lawful basis you can point to under UK GDPR and PECR.
One number is easier for clients and for records, and a shared inbox with assignment solves the routing problem that multiple numbers are usually trying to solve. Separate numbers make sense where the business lines genuinely have separate brands or separate regulatory permissions — see personal number versus brokerage number for the underlying trade-off.
The technical steps are short — deletion and registration are a matter of minutes, and coexistence onboarding requires synchronisation to complete within 24 hours. What sets the real timetable is Meta's side: display name review, business verification and template approvals all run on their own clock, which is why they should be started well before the day you intend to switch.
Shared WhatsApp inbox, client records, follow-ups and opportunities for the whole brokerage. 15-minute demo.
Two grey ticks are not proof of receipt. What sent, delivered, read and failed mean on the WhatsApp Business Platform, and what a broker should file instead.
Cloud API stores message data where you tell it to, but only if you set the region before registering the number. What EU and UK brokers should check.
Why clients see a number instead of your firm, how Meta reviews display names, and which of your legal, trading or network names belongs at the top of the thread.
