How to Auto-Reply on WhatsApp Using n8n

Every small business owner who runs support or sales over WhatsApp eventually hits the same wall: messages come in faster than you can answer them, and the ones that land at 11pm or on a Sunday just sit there until Monday. A whatsapp auto reply n8n workflow fixes that without hiring anyone or buying a bloated helpdesk suite. I have built a few of these for clients, and the setup is simpler than most people expect.

What You Actually Need

You need three things: a WhatsApp Business API connection (Meta's Cloud API or a provider like Twilio or 360dialog), an n8n instance (self hosted or n8n cloud), and a clear idea of what "auto reply" should mean for your business. That last part matters more than the tech. Are you confirming receipt of a message? Answering FAQs? Routing urgent messages to a human? Decide that first.

• WhatsApp Business API access, verified and approved
• An n8n workflow with a Webhook trigger listening for incoming messages
• A response node (HTTP Request or the WhatsApp node) that sends your reply back
• Optional: a condition node so not every message gets the same canned response

Building the Core Workflow

The basic whatsapp auto reply n8n setup is a webhook that catches the incoming message payload from WhatsApp, a Set or Function node that reads the message text, an IF node that checks for keywords or business hours, and an HTTP Request node that sends the reply through the WhatsApp API. That is the whole skeleton. Where it gets useful is the IF node: route "pricing" questions to one reply, "hours" to another, and anything you cannot classify to a message that says a human will follow up soon, then pings your team on Slack.

Where People Get It Wrong

Most people either overbuild this with a huge decision tree that breaks the moment a customer phrases something unexpectedly, or they underbuild it with one static reply that annoys people who ask a second question. The fix is keeping the automation narrow and honest. Auto reply what you can answer with confidence, and hand off everything else immediately instead of trying to fake a conversation.

The Esipick Take

We run something close to this internally at Esipick for initial client inquiries, and the lesson that stuck with me is that the auto reply's only real job is buying time honestly. It should acknowledge the message, set an expectation for when a real answer is coming, and stop there. The first version we built tried to answer too much and it showed, customers could tell they were talking to a script pretending to be a person. Once we scoped it down to "confirm and route," it started actually saving us time instead of creating cleanup work.

If you are staring at a WhatsApp inbox that never empties, this is a weekend project, not a quarter long one. Start with the webhook, wire up one or two reply branches, and expand from there once you see what people actually ask.