Integrating WhatsApp Business API with Salesforce: the real options

If your sales team lives in Salesforce and your customers live on WhatsApp, you have the classic problem: reps handle WhatsApp from their personal phones and the CRM stays empty. Nobody knows what was promised, there's no history on the opportunity, and when the rep leaves, the conversations leave with them.

Connecting the two can be done three ways. The decision isn't technical, it's about who carries the maintenance and who owns the WhatsApp account.

We're writing this from both sides: we work in Salesforce and we're an authorized Meta Tech Provider. It's an uncommon combination, which is why we see both classic failures, the CRM consultant who doesn't understand Meta's rules and the WhatsApp provider who doesn't understand the Salesforce data model.

Before choosing: two non-negotiable rules

The WhatsApp Business account stays in your company's name. Not the provider's, not the integrator's. Meta documents the structure in WhatsApp Business accounts. If the account belongs to someone else, the day you switch providers you lose the number, and the number is the asset. We develop this in official API vs unofficial solutions.

Meta charges per conversation, always. Whether it routes through Salesforce or anywhere else, Meta's messaging cost exists and it's separate from CRM licensing. It's in Meta's pricing model and we ground it in how much the API costs in Mexico. Anyone quoting you "WhatsApp in Salesforce" without mentioning this cost isn't quoting you the whole thing.

Option 1: Salesforce's native channel

Salesforce has first-party support for messaging channels, WhatsApp included, within its customer service stack.

For: conversations become native records, tied to the contact, the case and the opportunity. Routing, queues and reporting are Salesforce's own, with nothing to build. For a team already living in Service Cloud it's the cleanest experience.

Against: it requires additional messaging licenses, and per-user cost scales with the team. In operations with many agents it's the most expensive option. And automated flow flexibility is whatever the product gives you, no more.

When we'd choose it: a mid-size or large service team already on Service Cloud, with licensing budget, where what matters is the human agent working without leaving the CRM.

Option 2: an AppExchange connector

Third-party connectors bridge your WhatsApp account and Salesforce.

For: fast to implement, usually cheaper than native licensing, and many ship with a flow builder, template management and their own inbox already built.

Against: you depend on a third party for something critical. Check three things before signing: which Meta account your number ends up under, whether the connector gets updated when Meta changes the API, and what happens to your data if the vendor disappears. Also, most store conversations in their own cloud and mirror them into Salesforce, which means your history lives in two places.

When we'd choose it: small and mid-size operations that need to be running in weeks rather than months, with the caveat of verifying account ownership in writing.

Option 3: direct integration against the Cloud API

You connect Meta's official API to Salesforce with custom development: platform services to receive the webhooks, objects to model the conversation, and business logic in Apex or flows.

For: total control. You decide the data model, how conversations attach to opportunities and cases, what gets automated and what doesn't. No third-party messaging licenses and no middlemen. It's the only option that supports unusual business logic, and most real operations have unusual logic.

Against: it's development, and development gets maintained. When Meta changes the API, the adjustment is yours. It needs someone who understands both platforms, which is exactly the scarce profile.

When we'd choose it: when the sales process is the business differentiator, when volume is high enough that per-agent licensing doesn't add up, or when no packaged tool models what the operation actually needs.

What has to be solved in all three

Whichever option you pick, these always show up and are better decided up front:

• Templates. Anything sent outside the 24-hour window needs a Meta-approved template. Who writes them, who submits them for review and who responds to a rejection has to be defined. The approval guide explains why it isn't a minor formality.
• Contact identity. WhatsApp gives you a phone number. Salesforce has contacts, leads and accounts. How a number resolves to the right record, and what happens when none exists, is the first business rule to write down. Leave it to the default and you end up with duplicate leads.
• Who owns the conversation. If a bot handles first contact and then escalates, the handoff has to leave a trail in the CRM, not start from zero.
• The messaging limit. Messaging limits start low and rise with good behavior. A big campaign on day one is how you start with a poor quality rating.
• Consent. Stored in the CRM, with date and source. It's what protects you if someone reports you, and what avoids the blocked number scenario.

Our recommendation by size

• Up to 5 agents, standard process: AppExchange connector, with the Meta account in your name.
• Service team on Service Cloud, licensing budget: native channel.
• Custom sales process, high volume, or logic no tool models: direct integration.

What we don't recommend in any scenario is leaving reps handling customers from their personal WhatsApp while the CRM gets filled in by hand. It's the starting point of almost every project that reaches us and the most expensive of all, except the cost never appears on an invoice.

If you want to review which of the three applies to your case, we look at it across our Salesforce consulting service and our WhatsApp automation service. And if you're not clear on what a Salesforce consultant does, start there.