How to build a WhatsApp chatbot: auto-replies and flows

Customers no longer watch your business hours: they ask about prices at midnight, track orders on the weekend, and want to book an appointment over lunch. No team can reply instantly around the clock — but a WhatsApp chatbot can. The word "chatbot" may sound technical and robotic, but a well-built bot is neither cold nor complex; it answers routine questions in seconds and hands the rest to the right person.

This guide explains what a WhatsApp chatbot is, the work it takes over, the rule-based and AI-based types, how to build one step by step, and the mistakes to avoid.

What is a WhatsApp chatbot?

A WhatsApp chatbot is a digital assistant that automatically answers and routes messages coming in over WhatsApp. When a customer writes, the bot greets them instantly, answers frequently asked questions, gathers information when needed, and passes the topic to the right team. Technically, a chatbot runs on the WhatsApp Business API, so that infrastructure has to be in place first. We covered what the API is in our WhatsApp Business API guide.

The key point: a chatbot doesn't replace your team; it frees your team's time from repetitive work. Where a human touch is needed, a human steps back in.

What does a chatbot do? The main benefits

A well-designed bot lifts both customer satisfaction and operational efficiency:

• Instant replies: Wait time drops to zero; the customer gets a response in seconds.
• 24/7 access: Questions are answered even after hours, so requests don't slip away.
• Deflects FAQs: Questions like "What are your hours?" or "Where's my order?" get resolved before ever reaching the team.
• Qualifies leads: With a few questions, the bot understands the request and routes the prospect to the right rep.
• Captures after-hours interest: It handles a night-time request, gathers details, and presents it ready to the team in the morning.

A good chatbot answers the frequent questions for you, so you can spend time on the conversations that truly matter.

Two types of chatbot: rule-based and AI-based

Chatbots work with two basic approaches:

• Rule-based (menu/flow): Offers the customer options ("1- Order, 2- Appointment, 3- Agent") and proceeds by choice. It's predictable, easy to build, and more than enough for most businesses.
• AI-based (NLP): Understands free-typed sentences ("is there a 3pm slot tomorrow?") and replies naturally. It's more flexible and shines where questions are complex and varied. Combined with your WhatsApp messaging, its impact compounds.

The healthiest approach is often a mix of the two: start with a clear menu and use AI for free-form questions when needed.

Greeting + menuOrder statusBooking / FAQTalk to an agent

Before you build: what will you automate?

A successful chatbot starts not with technology but with the right question: "Which questions does my team answer over and over?" Look at the last week or two of conversations and list the 5-10 most common requests. Usually most of them cluster into a few topics: price/quote, order or shipping status, hours/location, booking, and "I'd like to speak to someone." Building your bot for these high-volume, repeating topics first delivers the fastest return.

How to build a WhatsApp chatbot, step by step

Done in the right order, the build is quite clear:

• 1. Prepare the API: A chatbot runs on the WhatsApp Business API; that infrastructure must be connected first.
• 2. List the questions: Nail down the 5-10 most common questions to automate and their answers.
• 3. Design the greeting and menu: Create a short greeting and a menu of 3-5 clear options.
• 4. Build the scenario flows: Define the steps for each option (question → answer → next step).
• 5. Add a human handoff point: Put in an exit that smoothly transfers the chat to an agent when the bot can't help.
• 6. Test and launch: Try it with real scenarios, fix where it stalls, then go live.

What makes a good greeting and menu?

The first message sets the tone of the experience. A good greeting is short, says clearly what it can do, and doesn't overwhelm. Three to five options is ideal; a ten-option menu is worse than no menu. Set expectations too: "Happy to help right away; you can type 'agent' anytime to reach our team." That single line keeps users from feeling trapped.

Example scenarios

The most-used flows that create quick value are:

• Order/shipping status: The bot asks for the order number and returns the status; it greatly relieves the support line.
• Booking: It offers available times, takes the choice, and sets a reminder. Strong alongside appointment reminder automation.
• FAQ: Instantly answers hours, location, returns and similar questions.
• Quote/price request: Understands the need with a few questions, qualifies the prospect, and routes to the right rep.
• Talk to an agent: An always-available "hand off to a human" option.

Why is the human handoff essential?

No bot can solve everything, nor should it try. A chatbot's most critical feature is being able to transfer the chat smoothly to a human when it reaches its limit. A bot that leaves users in an endless menu loop is the most complained-about bot. Good design can say "I couldn't help with that, let me connect you to an agent" and carry the context gathered so far to that agent during the handoff.

Common mistakes

• Too robotic: Cold, canned sentences lower trust; use a warm, plain voice.
• No exit to a human: Trapping the user inside the bot is the biggest mistake.
• Too many options: Ten-item menus confuse; simplify.
• Dead ends: What happens when the bot doesn't understand (a fallback) must be defined.
• Launching without testing: A bot untested on real scenarios stalls on the first customer.

Measuring chatbot performance

The work doesn't end once the bot is live; improvement comes from data. Track these metrics: the share the bot resolves without ever reaching the team (deflection), the completed-flow rate, how often it hands off to a human, and customer satisfaction. A high handoff rate means the menu or flows fall short; a low completion rate means users get stuck somewhere. The most practical way to keep these metrics with your conversations is to bring multichannel communication into one hub.

Does a small business need a chatbot?

Yes — and small teams are often the ones who benefit most. Without a large support department, answering repetitive questions eats your valuable time. A chatbot works like an assistant that "greets the customer while you're busy," even in a one-person business: it answers common questions, collects the request, and you take over when you're free. It makes sense financially too; a single missed sale often exceeds the monthly cost.

Chatbot, live support, and what gets confused

These terms are often mixed up. A chatbot replies automatically; live support is a real agent chatting in real time. The two aren't rivals but complements: the bot handles the first touch and frequent questions, then hands off to live support (a human) when needed. The best experience is one where the customer clearly knows when they're talking to the bot versus a human, and the transition is seamless.

Taking the chatbot to other channels

You don't have to keep the same logic only on WhatsApp. A well-built flow can also run in the chat box on your website, on Instagram, or other channels. What matters is that the replies from all these channels are collected in one place and the customer history stays unified — so even if the customer switches channels, the conversation picks up where it left off.

WhatsApp chatbot with Rocketly

With Rocketly you design your WhatsApp chatbot in a visual flow builder, no code required, and turn on AI-assisted replies if you wish. When the bot can't help, the chat is handed to your team with all the context gathered so far, and every conversation is logged automatically to the CRM. You carry the requests you collect to a close with proper lead tracking and never miss after-hours opportunities.