By Quick answer: How do I build a chatbot for my business?
- Pick one job — FAQs, bookings, order status or lead qualification.
- Pick the channel — website widget, inside your app, or WhatsApp (where most Indian customers already are).
- Pick the brain — rule-based buttons, an AI bot trained on your documents, or a hybrid.
- Always hand off to a human when the bot is unsure.
- Measure resolution rate, handoff rate and customer satisfaction.
Key takeaways
- Start with the 20 questions that make up most of your support volume. Automate those well before anything else.
- Hybrid bots — buttons for structured tasks, AI for open questions — usually beat either alone.
- WhatsApp is the natural channel in India; the WhatsApp Business Platform charges per delivered template message by category.
- An AI bot must answer from your approved content and say 'I don't know' instead of guessing.
- A clear path to a human protects customers and your brand.
Every business has the same twenty questions arriving over and over: timings, prices, delivery status, "do you have this in size M?", "can I reschedule?". A good chatbot answers those instantly at 2 a.m. and passes everything else to a person. A bad one traps customers in loops and costs you their goodwill. The difference is mostly design, not technology.
Step 1: choose the job
Pick one job and do it well before adding more:
Answer FAQs
Timings, policies, pricing, how-tos. The easiest win; start with your 20 most common questions.
Bookings and orders
Appointments, reservations, order status, reschedules. Needs integration with your calendar or order system.
Lead qualification
Ask a few questions, capture contact details, route hot leads to sales.
Internal helpdesk
HR and IT questions for employees, answered from your policy documents.
Step 2: choose the channel
| Channel | Best for | Notes |
|---|---|---|
| Customer service and updates in India | Uses the WhatsApp Business Platform; Meta charges per delivered template message by category and country12 | |
| Website widget | Pre-sales questions and lead capture | Easy to add; visitors expect quick, specific answers |
| Inside your app | Support for logged-in users | Can use account context — orders, plans, history |
| Instagram / Messenger | Brands selling through social | Same Meta platform family |
In India, WhatsApp is usually where customers already are. Meta's pricing is per delivered message rather than per conversation, varies by message category — marketing, utility, authentication and service — and by country, with India rate cards published in rupees1.
Step 3: choose the brain
Rule-based (buttons and flows)
- Predictable, easy to approve and test
- Great for structured tasks: book, track, pay
- Cheap to run
- But handles only what you designed
AI (a language model on your content)
- Understands free-text questions in many phrasings
- Answers from your documents with retrieval
- But needs guardrails, testing and monitoring
- And carries a running cost per message
Most businesses do best with a hybrid: buttons for structured tasks, AI for open questions, and a human for everything else.
Step 4: build it
- Collect your content — FAQs, policies, product info, past support replies. Clean and update it first.
- Design the flows for structured tasks, with clear buttons and a visible "talk to a person" option.
- Set up the AI part with retrieval over your approved content; instruct it to answer only from that content and to say when it does not know.
- Integrate with your calendar, order system, CRM or helpdesk so the bot can actually do things.
- Design the handoff — when the bot is unsure, the customer is upset, or the topic is sensitive, pass the chat to a person with the full context.
- Test with 100 real questions from past chats, including awkward and off-topic ones.
Step 5: protect customers and your brand
- The bot never invents prices, policies or promises — it answers only from approved content
- Users can always reach a human
- No sensitive data is requested in chat that does not need it
- Prompt-injection and abuse are considered for AI bots (see the OWASP Top 10 for LLM apps3)
- A privacy notice explains what is stored and for how long, in line with India's DPDP Rules4
- Marketing messages are sent only to customers who opted in
Step 6: measure and improve
| Metric | What it tells you |
|---|---|
| Resolution (containment) rate | Share of conversations the bot completed without a human |
| Handoff rate | How often it needs help — and on which topics |
| Customer satisfaction | Whether resolved actually felt resolved |
| Unanswered questions | Your content gaps, ranked |
| Business outcome | Tickets reduced, bookings made, leads qualified |
Review failed conversations every week. They are the most useful product research you will get.
What it costs
A rule-based bot on a no-code platform can cost a modest monthly subscription. A custom AI chatbot trained on your content, integrated with your systems and with human handoff typically costs ₹3–15 lakh to build in India, plus model usage and, on WhatsApp, Meta's per-message charges. For a full assistant product, see how to make an app like ChatGPT.
Key terms, defined
Chatbot — software that converses with users in text or voice. Rule-based bot — follows designed flows. AI chatbot — uses a language model to understand and answer. Handoff — passing a conversation to a human. Template message — a pre-approved WhatsApp message a business can send outside an open conversation. Containment rate — the share of chats resolved without a human.
Frequently asked questions
How much does it cost to build a chatbot for a business in India?
A rule-based website or WhatsApp bot built on a no-code platform can cost a small monthly subscription. A custom AI chatbot trained on your documents with integrations and human handoff typically costs ₹3–15 lakh to build, plus model usage and, for WhatsApp, Meta's per-message charges.
What is the difference between a rule-based and an AI chatbot?
A rule-based chatbot follows designed flows with buttons and keywords — predictable but limited. An AI chatbot uses a language model to understand free text and answer from your content — flexible but needs guardrails. Many businesses combine both.
How do I build a WhatsApp chatbot?
Get access to the WhatsApp Business Platform through Meta's Cloud API or a business solution provider, verify your business, register a number, create approved message templates, then connect a bot platform or your own backend that receives messages and replies. Meta charges per delivered template message, by category and country.
Can a chatbot answer questions from my own documents?
Yes. AI chatbots use retrieval-augmented generation: your documents are indexed, and for each question the most relevant passages are retrieved and given to the model, which answers from them. Instruct it to cite sources and to say when the answer is not in your content.
Will a chatbot replace my support team?
It should handle repetitive questions so your team can focus on complex ones. Most businesses see the best results when the bot resolves routine queries and hands everything else to people with the conversation context attached.
How do I measure if my chatbot is working?
Track the resolution (containment) rate, the handoff rate, customer satisfaction after chats, the questions it failed to answer, and the business outcome — tickets reduced, bookings made or leads qualified. Review failed conversations weekly.
Sources
- Meta for Developers — Pricing on the WhatsApp Business Platform
- WhatsApp Business — Platform pricing (per-message model and categories)
- OWASP — Top 10 for Large Language Model Applications
- Press Information Bureau — Government notifies DPDP Rules (2025)
- Checked October 23, 2026. Figures that change — fees, salaries, prices — are reviewed every six months; if something is out of date, tell us.
Where to go from here
- How to make an app like ChatGPT — when you need a full AI assistant
- RAG explained like you're five — how bots answer from your documents
- How to make an appHow to make an app like ChatGPTBuild a ChatGPT-style app on a model API: streaming chat, conversation memory, your own data, guardrails, usage limits and costs — with a realistic build plan.
- How to make an appHow to make an app with zero coding experience (2027 guide)You can make an app without coding in 2027: validate the idea, design it, then build with no-code tools, AI app builders or a hired developer. Step by step.
- Career guidesAI customer service jobs in 2027: designing chatbot flows, human handoffs, bot quality checks, guardrails, transparency, measuring automation and the new roles around AI agentsAI customer service jobs in 2027: chatbot flows, human handoffs, bot quality checks, guardrails, transparency, automation metrics and support pay in India.
- How to make an appHow to make an app for free: what 'free' really costsYou can build an app for ₹0 with free tiers, but publishing and growing cost money. The free tools, their real limits, and the costs that appear later.
- App developmentAI app development cost in 2027: what you'll actually payAn AI app costs ₹12–60 lakh to build in India, ₹5–20 lakh to add AI to an existing app — plus a cost on every request. The full build and per-user running math.
- How to make an appHow to make an Android app from scratchMake your first Android app with Android Studio, Kotlin and Jetpack Compose: setup, first screen, state, data, testing on a phone and publishing on Google Play.
Get the ideas, not the noise.
Sign up with Google and we’ll send you short UX psychology tips by email — why people tap, stall and leave, and what to do about it. Unsubscribe anytime. Prefer something to keep? Get the free Laws of UX guide as a PDF.
Make sure the first screen makes sense.
A behavioral UX audit of your app or SaaS product: why users hesitate, stall and leave, and a prioritized fix list. Fixed price, 2–4 weeks.