Automate WhatsApp with n8n: Replies, Leads and Alerts

In LATAM, customers don't fill out forms: they message you on WhatsApp, and whoever replies first wins the sale. Automating WhatsApp with n8n means connecting Meta's WhatsApp Business Cloud API (the official one, free to access) to workflows that answer common questions instantly, create every lead in your CRM with its conversation, and send automatic order and payment alerts. n8n ships native nodes for this: there's no need to build an application from scratch. This guide covers the full path, including what almost nobody tells you about the risks of unofficial solutions.

Official API vs unofficial solutions: the decision that defines everything

Before building your first workflow, you have to choose how to connect to WhatsApp, and this is where the expensive mistakes happen.

Criterion	Meta Cloud API (official)	Unofficial APIs (emulate WhatsApp Web)
Legality with Meta	Allowed, with support	Violates the terms of service
Ban risk	Low with good practices	High and without prior warning
Access cost	Free (you pay per conversation)	USD 20-100/month license
Cold outbound messages	Approved templates only	No restriction (which is why they get banned)
Stability	High, Meta's infrastructure	Breaks with every WhatsApp update
Verified number and business name	Yes	No

Unofficial tools are tempting because they let you send anything to anyone without approval. That's exactly why Meta hunts them down: if your years-old business number gets banned, no support can recover it. We've seen businesses lose their main sales channel overnight. For personal experiments, it's your call; for a company, the answer is the official API, always.

The one thing the official API won't allow? Messaging cold anyone who didn't contact you first, except with templates approved by Meta and prior consent. For a serious business, that isn't a limitation: it's the difference between marketing and spam.

What you need before you start

1. A verified Meta Business account (free, takes anywhere from hours to a few days).
2. A phone number that isn't in use on personal WhatsApp (it can be a new line or a landline).
3. An app on Meta for Developers with the WhatsApp product enabled: that's where you get the access token and the number ID.
4. An n8n instance reachable over HTTPS, because Meta delivers incoming messages via webhook and needs a public URL. If you don't have one yet, follow our self-hosted install guide with Docker first.

With that, the connection in n8n is two nodes: WhatsApp Trigger (receives messages) and WhatsApp Business Cloud (sends replies, templates, images and documents).

Case 1: automatic replies to frequent questions

The simplest flow and the one that relieves the team the most: between 40% and 70% of a business's incoming questions are always the same ones (hours, prices, shipping, payment methods).

The workflow: WhatsApp Trigger → a Switch node that classifies the question → matching reply → if nothing matches, hand off to a human. Classification can be by keywords (fast and free) or with an AI model that understands natural language — the difference between "reply if they type 'price'" and "understands that 'how much is shipping to Córdoba?' is a shipping question". That second version is what we build in AI chatbot projects, where the bot also checks real stock and prices against your systems.

Golden rule: always leave a path to a human. A bot that traps the customer in a loop creates more anger than having no bot at all.

Case 2: WhatsApp leads straight into the CRM

The case with the highest measurable return. Without automation, WhatsApp leads live on someone's phone: they aren't logged, assigned or followed up. The workflow that fixes it:

1. WhatsApp Trigger: a message comes in from a new number.
2. CRM lookup: does the contact exist? If not, it's created with name and phone.
3. Qualification: an AI node extracts what the question is about (product, urgency, area).
4. Assignment: the rep is assigned per business rules and notified with full context.
5. Logging: the conversation is linked to the contact.

A real estate agency we work with went from answering inquiries in 2-4 hours to confirming receipt in seconds and assigning the right agent in under a minute. In markets where five competitors get the same inquiry, that delta is the difference between showing the property or finding out it was already seen with someone else. If your current CRM has no API or doesn't fit your process, we also handle that part with custom CRMs.

How many WhatsApp leads do you lose every month without logging them? Book an intro meeting and we'll show you the flow running on a case similar to yours.

Case 3: automatic order and payment alerts

Here WhatsApp stops being an inbound channel and becomes a notification channel: "your order is confirmed", "your payment cleared", "your appointment is tomorrow at 10". These messages use utility templates approved by Meta, which are the cheapest (cents of a dollar, or free inside the service window).

The trigger is no longer WhatsApp but your system: a webhook from your store, your management system or your payment gateway fires the workflow, n8n builds the message with the order data and sends it. The typical impact: a 30-50% reduction in "did my order ship?" questions, which are pure operational cost. If you collect payments with Mercado Pago, this flow pairs very well with collections automation with n8n.

Example workflow in JSON

This is the skeleton of a real auto-reply workflow with handoff. You can import it into n8n (menu → Import from File) and replace the credentials:

{
  "name": "WhatsApp - Respuesta y derivación de leads",
  "nodes": [
    {
      "name": "WhatsApp Trigger",
      "type": "n8n-nodes-base.whatsAppTrigger",
      "parameters": { "updates": ["messages"] },
      "position": [0, 0]
    },
    {
      "name": "Clasificar consulta",
      "type": "n8n-nodes-base.switch",
      "parameters": {
        "rules": {
          "rules": [
            { "operation": "contains", "value2": "precio" },
            { "operation": "contains", "value2": "envío" },
            { "operation": "contains", "value2": "horario" }
          ]
        }
      },
      "position": [220, 0]
    },
    {
      "name": "Responder precios",
      "type": "n8n-nodes-base.whatsApp",
      "parameters": {
        "operation": "send",
        "textBody": "¡Hola! Te paso nuestra lista de precios actualizada: {{LINK}}. ¿Buscás algo en particular?"
      },
      "position": [460, -120]
    },
    {
      "name": "Derivar a humano",
      "type": "n8n-nodes-base.whatsApp",
      "parameters": {
        "operation": "send",
        "textBody": "¡Gracias por escribirnos! En menos de 15 minutos te responde una persona del equipo."
      },
      "position": [460, 160]
    },
    {
      "name": "Notificar al equipo",
      "type": "n8n-nodes-base.slack",
      "parameters": { "channel": "#leads-whatsapp" },
      "position": [680, 160]
    }
  ]
}

It's deliberately simple: the production version adds error handling, the 24-hour window, CRM logging and AI classification. But it's enough for you to see the structure and try it today.

When automating WhatsApp does NOT pay off

• If you have fewer than ~20 conversations a day, the cost of building and maintaining this outweighs the savings: answer by hand and reassess when volume grows.
• If your sale is 100% consultative and high-ticket, automate only intake and logging; leave the sales conversation to people.
• If you were planning to use an unofficial API to blast cold mass promos: don't. Today's savings are tomorrow's banned number.
• If your internal process is a mess, the bot will answer fast… the wrong things. Tidy up first, then automate — we explain it in how n8n works.

Build it yourself, or have it built turnkey

Everything you've read can be done at home with time and patience: Meta account, n8n instance, approved templates, testing. Or it can be done in a couple of weeks with people who've built it many times. At Deepyze we implement the full integration —Meta's official API, workflows with error handling, connection to your CRM and AI bots that query your real systems— as part of our AI automation service. Tell us about your case and within 24 hours you'll have a fixed-price proposal, from a team in your time zone that also answers on WhatsApp.