How to Build a WhatsApp Chatbot — Complete Setup Guide

What is a WhatsApp chatbot and how does it work?

A WhatsApp chatbot is an automated program that handles customer conversations on your WhatsApp Business number. When a customer messages you, the bot intercepts the message, determines what the customer needs, and responds automatically — without a human involved.

Chatbots on WhatsApp work through the Business API. When a message arrives at your API number, it's routed to your platform (WA.Expert), which runs it through your bot logic before sending a response back via the API to the customer.

Do you need coding skills to build a WhatsApp chatbot?

No — not with a modern chatbot platform. WA.Expert's chatbot builder is entirely visual. You design conversation flows by dragging and dropping message nodes, connecting them with arrows, and configuring responses through a form-based interface.

You do need coding skills only if you want to:

• Build custom integrations with internal systems not covered by native connectors
• Create highly dynamic responses that pull real-time data from your own database via API calls
• Build a fully AI-powered bot using GPT or Claude that requires custom prompt engineering

For 90% of business use cases — FAQ handling, lead qualification, appointment booking, order status, basic support — zero coding is required.

What can a WhatsApp chatbot do vs what it cannot do?

What is a chatbot flow and how do you design one?

A chatbot flow is the decision tree that maps out every possible path a conversation can take. Think of it as a flowchart: customer sends X → bot responds with Y → customer picks option A or B → bot follows path A or path B.

Designing a good flow starts with mapping your most common customer conversations. Look at your WhatsApp chat history and identify the top 10 questions customers ask. Then design a flow that:

1. Greets the customer with a welcome message and menu of options
2. Routes based on their selection to topic-specific sub-flows
3. Asks clarifying questions when needed to narrow down the response
4. Delivers the answer, confirmation, or resolution
5. Offers a next step (book, buy, speak to a human)

The best chatbot flows are short. Customers abandon long bot conversations. Aim for 3–5 exchanges maximum before delivering value or routing to a human.

What are keyword triggers and how do you set them up?

Keyword triggers are words or phrases that automatically activate a specific bot response or flow. When a customer sends a message containing a trigger word, the bot recognises it and responds with the corresponding flow.

Examples of keyword triggers:

• "price" or "pricing" → triggers your pricing information flow
• "order" or "track" → triggers your order tracking flow
• "human" or "agent" or "support" → triggers human handoff
• "hi" or "hello" → triggers your welcome message
• "refund" or "return" → triggers your returns policy flow

In WA.Expert's chatbot builder, keyword triggers support exact match, contains, and regex matching. You can assign multiple keywords to the same flow and set fallback responses for messages that don't match any trigger.

How do you hand off from bot to a human agent?

Human handoff is one of the most important elements of a WhatsApp chatbot. A bot that traps customers in loops without an escape route creates frustration and damages your brand.

Handoff triggers in a well-designed bot:

• Customer request: Any message containing "human," "agent," "speak to someone," or similar keywords immediately routes to a human
• Bot failure: If the bot can't match a customer's message after 2–3 attempts, it automatically escalates
• Flow completion: After the bot resolves the query, it can offer "Was this helpful? / Speak to a human" as an option
• Escalation rules: Specific topics (complaints, refunds, legal matters) automatically bypass the bot and go straight to a human
• Time-based: During business hours, some businesses configure the bot to offer immediate human connection; outside hours, it collects the query for a human to follow up

In WA.Expert, when a handoff is triggered, the conversation appears in the team inbox and can be assigned to a specific agent or a queue. The agent sees the full bot conversation history for context.

How do you connect a chatbot to your CRM or database?

Chatbot-CRM integration means the bot can look up customer information, create records, and update data in real time during a conversation. This enables powerful use cases like: bot looks up order status by customer phone number, bot creates a lead record when a new prospect messages, bot qualifies a lead and passes the score to your CRM.

WA.Expert's native CRM integrations include Zoho, HubSpot, Google Sheets, and Shopify. For other systems, webhook-based integrations allow the bot to make HTTP calls to any REST API — enabling connection to virtually any database or CRM.

How much does it cost to run a WhatsApp chatbot?

Running a WhatsApp chatbot has two cost components:

• Platform subscription: WA.Expert's monthly fee covers the chatbot builder, team inbox, and all platform features
• Meta conversation charges: When the bot responds to a customer-initiated message, it's a service conversation (~₹0.29 per 24-hour window in India). The first 1,000 service conversations per month are free. There is no per-message charge — all messages within the open window are covered by one conversation charge.

For most businesses, the Meta conversation cost is minimal because inbound service conversations are cheap and partially covered by the free tier. The main cost is the platform subscription.

How do you measure if your chatbot is working?

A chatbot that looks good in demos but doesn't help customers is a waste. Here are the metrics that matter:

• Containment rate: % of conversations fully resolved by the bot without human involvement. Target: 60–80% for FAQ-heavy use cases.
• Fallback rate: % of messages where the bot couldn't find a matching response. High fallback = your flows don't cover real customer needs.
• Handoff rate: % of conversations escalated to a human. Context: some handoffs are healthy (complex queries), but a rate above 40–50% suggests the bot isn't handling enough.
• Response time: Time from customer message to bot response. Should be under 3 seconds.
• Customer satisfaction: Post-conversation rating (if you add a rating step to your flow). Compare bot vs human CSAT.
• Conversion rate: For lead-gen bots — % of bot conversations that result in a qualified lead or booking.

How to build your WhatsApp chatbot — step by step

1. Sign up for WA.Expert and connect your WhatsApp numberStart your free trial, connect your existing WhatsApp Business number or register a new one. Setup takes under 24 hours.
2. Map your top 10 customer queriesCheck your existing WhatsApp chats or customer service records. List the questions customers ask most often — these become your bot's core flows.
3. Build your welcome flowDesign the greeting message and main menu. Keep it to 3–5 options maximum. Every option should link to a sub-flow.
4. Create sub-flows for each topicBuild a separate flow for each menu option — pricing, support, booking, FAQ, etc. Each sub-flow should resolve in 3–5 exchanges.
5. Set up keyword triggersAdd keywords that map to each flow. Include common misspellings and variations. Test thoroughly.
6. Configure human handoffSet up escalation triggers — keywords, failure conditions, and specific topic routes. Test that handoff works correctly before going live.
7. Test with real messagesUse WA.Expert's test mode to send messages and verify responses. Test edge cases — what happens when a customer sends something unexpected?
8. Go live and monitor the first week closelyWatch your fallback rate in the first week. Every failed bot response is a gap to fill with a new flow or keyword trigger.