A WhatsApp auto-reply system lets a business respond to every customer inquiry immediately, classify it by topic, and move it forward to booking an appointment or handing off to a human agent. Instead of waiting for a reply during business hours, the customer gets an instant response, and you get the inquiry organised and logged in your system.
We build WhatsApp auto-reply systems on top of the WhatsApp Cloud API, integrated with CRM, calendar, and n8n workflows. Every incoming message enters a defined process: acknowledgement message, clarifying questions, routing to the right agent, or booking an appointment directly from the conversation.
Why WhatsApp auto-reply and not just an away message?
A WhatsApp auto-reply system differs from a regular away message in that it responds to the content of the message, not just notifies that you’re unavailable. An away message appears once and says “We’ll get back to you soon.” An auto-reply system asks the customer what they need, offers options, and moves them to the next step.
When a customer sends “I want to book an appointment”, an automated system can ask which type of service, offer available dates from your calendar, and confirm the appointment. When they ask “How much does it cost”, the system can present a price list or ask additional questions to understand which product fits. All of this happens in real time, even at ten o’clock at night.
We integrate the auto-reply with chatbots that handle the conversation from start to finish, including handoff to a human agent when needed.
How do you build an auto-reply that actually works?
A WhatsApp auto-reply system is built on three components: the WhatsApp Cloud API that connects your business number, an n8n workflow that defines the conversation logic, and an integration to CRM or Google Sheets that logs the inquiries.
The process starts with registering for the WhatsApp Cloud API through Meta Business. After number approval, we configure a webhook that passes every incoming message to an n8n workflow. The workflow checks the message content, identifies intent (“book appointment”, “ask question”, “get price”), and sends an appropriate response.
Each response can be text, buttons for selection, or a list of options. If the customer chooses “book appointment”, the system pulls availability from your calendar, displays available dates, and updates the calendar when the customer confirms. If they ask a question the system doesn’t recognise, it hands off the conversation to an agent and preserves all the history.
| Component | Role | Example |
|---|---|---|
| WhatsApp Cloud API | Connects the business number and passes messages | Receives “Hi, I want to book an appointment” |
| n8n workflow | Analyses the message and decides how to respond | Identifies “book appointment” and pulls availability |
| CRM integration | Saves the inquiry and updates the customer in the system | Creates a new lead with customer details |
| Calendar | Shows availability and updates appointments | Shows “Tuesday 14:00 or Wednesday 10:00” |
We build all of this as a single system, with monitoring that alerts if a message gets stuck or if the API is unavailable.
How long does it take to set this up?
Setting up a WhatsApp auto-reply system takes between two weeks and a month, depending on the complexity of the processes you want to build. The first stage is approving the business number in the WhatsApp Cloud API, which typically takes three business days after submitting documents.
After approval, we build the workflow: define conversation scenarios, write the messages, and connect the integrations to CRM and calendar. This takes one to two weeks, depending on the number of scenarios. If you only want initial response and routing to an agent, it can be ready in a week. If you want appointment booking, inventory lookup, and integration with accounting software, it takes longer.
After launch, we go through a two-week calibration period: check what questions customers actually ask, add answers to questions we didn’t anticipate, and improve the phrasing. A system that has been calibrated answers 70-80% of inquiries without human intervention.
What happens when the system doesn’t understand the customer?
When a WhatsApp auto-reply system doesn’t recognise the customer’s intent, it needs to say so clearly and hand off to a human agent. We build a clear fallback: if the customer sent a message the system doesn’t recognise, or if they wrote “I want to speak to a person”, the conversation transfers to an agent and preserves all the history.
Handoff to an agent can be immediate during business hours, or send an alert in Slack or email outside business hours. The agent sees the entire conversation so far, including what the customer requested and what the system already answered. This saves them from having to ask again “How can I help you”.
Some businesses prefer the system to try to understand once more before handoff. In that case, we define a clarification question: “I didn’t quite understand. Are you interested in booking an appointment, product information, or something else?” with selection buttons. Only if the customer doesn’t respond or chooses “other” does the system hand off to an agent.
How do you connect this to CRM and calendar?
Connecting a WhatsApp auto-reply system to CRM and calendar is done through API or webhooks, depending on the system you’re using. When a customer sends a first message, the n8n workflow checks if their number already exists in the CRM. If yes, it pulls their details and history. If not, it creates a new lead.
Every message is logged in the CRM as an interaction. If the customer requested to book an appointment, the system pulls availability from your calendar through Google Calendar API or Outlook API, shows them available dates, and updates the calendar when they choose. The appointment is created with all details: customer name, phone number, service type, and notes.
If you’re using accounting software like Green Invoice, the system can also send a quote or invoice directly in the conversation. The customer receives a link, views the document, and can approve or ask questions. All of this is logged in the CRM so you see the entire process in one place.
For those starting out without a CRM, we can build the system on top of Google Sheets. Every inquiry enters as a row in the sheet, with date, time, phone number, message content, and status. It’s less sophisticated than a full CRM, but it works and you can upgrade later.
How much does it cost to run WhatsApp auto-reply?
The cost of running a WhatsApp auto-reply system comprises three components: the cost of conversations in the WhatsApp Cloud API, the cost of the platform running the workflow (n8n or Make), and the cost of building and implementation.
The WhatsApp Cloud API charges per conversation, not per message. A conversation is a 24-hour window from the moment you send a message to the customer or they send to you. All messages within those 24 hours count as one conversation. The price depends on conversation type: a conversation the customer opened (user-initiated) is cheaper than one you opened (business-initiated). A small business with a hundred conversations a month will typically pay a few dozen dollars.
n8n, the platform we build workflows on, offers a free plan up to 2,500 executions per month. A business receiving dozens of inquiries per day can start with the free plan. When volume grows, the paid plan starts at $20 per month.
Building cost depends on complexity: a basic reply system with routing to an agent can be ready in a few working days. A full system with appointment booking, CRM integration, and multiple scenarios takes longer. We work on a project basis or ongoing support, depending on what fits the business.
What’s the difference between auto-reply and a full chatbot?
A WhatsApp auto-reply system is part of a chatbot, but not every auto-reply is a full chatbot. Auto-reply responds to incoming inquiries according to a defined template: the customer sends a message, the system identifies a topic and responds. A full chatbot can also initiate conversations, send reminders, follow up with customers who didn’t complete a process, and manage a complex conversation with context.
For example: auto-reply will tell the customer “We received your inquiry, we’ll get back to you soon.” A full chatbot will ask them additional questions, check if they’re an existing customer, offer them relevant products based on their purchase history, and send them a reminder if they haven’t responded within two days.
Most businesses start with auto-reply and expand gradually. After the system is working and you see which questions recur, you can add capabilities: appointment reminders, quote follow-up, order status updates. Each such addition turns the system more into a chatbot and less just a reply.
We build both approaches, depending on what the business needs now and where it wants to go. A good system is one that starts simple, works from day one, and can be expanded without rebuilding everything.
How do you measure if the system is working?
Measuring success of a WhatsApp auto-reply system starts with a simple question: how many inquiries did the system close without human intervention. We call this “automatic resolution rate”. A good system should close 60-70% of inquiries on its own in the first stage, and 80% and above after calibration.
The second metric is first response time. Before automation, a customer waits on average several hours until an agent responds. With auto-reply, the response comes within seconds. This doesn’t just improve customer experience, it also increases the chance they’ll stay in the conversation and not turn to a competitor.
The third metric is the number of appointments or leads created through the system. If we built an appointment booking process, we count how many appointments were booked automatically versus how many required intervention. If we built a lead capture process, we check how many leads the system logged in the CRM and how many of them became customers.
All these metrics are measured directly from the n8n workflow and from the CRM. We build a dashboard that shows these numbers in real time, so you see what’s working and where improvement is needed. If you want to understand more about CRM lead automation, we have a separate guide on that.
Can you use the system for outbound messages too?
Yes, but with restrictions. The WhatsApp Cloud API allows you to send outbound messages to customers who have contacted you in the past, but only within specific windows and with approved message templates. You can’t send marketing messages freely like email.
An outbound message must use a pre-approved template that Meta has reviewed. Common templates include appointment reminders, order confirmations, delivery updates, and payment reminders. Each template must be submitted for approval before use, and approval typically takes one to two days.
Once a customer responds to your outbound message, a 24-hour conversation window opens and you can send free-form messages. This is useful for appointment confirmations: you send a reminder template (“Your appointment is tomorrow at 14:00”), the customer replies “confirmed”, and you can continue the conversation naturally.
We build outbound messaging as part of the overall system when it makes sense: appointment reminders from the calendar, payment reminders from the invoicing system, or order updates from the inventory system. Each outbound message is logged in the CRM alongside the rest of the conversation history, so you have a complete view of every customer interaction.
Frequently asked questions
Can you use WhatsApp auto-reply with a regular account?
No. Auto-reply requires the WhatsApp Business API, not the regular app or WhatsApp Business App. The API allows connecting external systems and sending automated messages, but it requires approval from Meta and technical setup. A regular account is limited to one away message and cannot run automated processes.
What happens if the customer writes in a language the system doesn't know?
It depends on how you built the system. If you're identifying intent by keywords, a message in another language won't be recognised and the system will hand off to an agent. If you're using a language model (GPT or Claude), the system can understand intent even in another language and respond in Hebrew or English. We recommend defining in advance which languages you support and handing off to an agent if the customer writes in another language.
How long does it take for the customer to get a first response?
The first response arrives within seconds from the moment the customer sends a message. The webhook passes the message to n8n, the workflow processes it and sends a response, and it all happens in real time. If there's load or if WhatsApp's API is slow, it can take up to 10 seconds, but that's rare. If the response takes more than a minute, something isn't working and we get an alert.
Can the system send files or images to the customer?
Yes. The WhatsApp Cloud API supports sending images, PDF documents, audio files, and videos. We use this to send price lists, quotes, contracts for signature, or usage instructions. The file can be stored in Google Drive or on your server, and the system sends a link or the file itself. There's a size limit of 100MB per file, but for most business content that's sufficient.