How much does a WhatsApp chatbot cost in India? An honest 2026 breakdown
Most “WhatsApp chatbot pricing” pages you’ll find online quote a build cost and stop. That’s only one of three layers you’re actually paying. Here’s the full picture, so you can structure a quote conversation with any vendor.
There are three layers of cost
- The build — one-time cost to design, code, integrate, and launch.
- The infrastructure — recurring monthly cost of hosting, monitoring, and the Business Solution Provider (BSP) you route through.
- WhatsApp’s conversation charges — Meta charges per 24-hour conversation, with rates set by category and country.
The cheapest of the three to underestimate is the third. Most teams realise it after launch, when a marketing burst pushes their Meta bill past their build cost.
1. Build cost
For a custom build delivered by a small product team in India, scope typically falls into three buckets:
- Starter: 5–10 intents, scripted flows, WhatsApp Business API setup, basic CRM webhook. 2–3 week build. Good for a single use-case — order status, lead capture, or appointment booking.
- Production: Intent classification, multi-turn flows, full CRM and billing integration, human handoff with context, analytics dashboard. 5–8 week build. The default for any team handling more than a few hundred inbound a month.
- Enterprise: Custom NLU model, multi-language, multi-region, A/B testing, dedicated SLA, audit logging. Quoted per scope.
If you go through a no-code platform (Wati, AiSensy, Gallabox), there’s no upfront build cost — you pay a monthly subscription instead. That’s usually cheaper for a year or two, then more expensive forever.
2. Infrastructure cost
Recurring monthly:
- BSP fee: If you go through a Business Solution Provider (Twilio, Gupshup, Karix, 360dialog, Wati), there’s a monthly fee that scales with volume. Going direct via Meta’s Cloud API has no BSP fee but more setup work.
- Hosting: A small bot on shared or serverless infrastructure is inexpensive; a dedicated database, queue, and observability stack costs more.
- LLM costs (if applicable): If your bot uses an LLM for intent classification or replies, expect a small per-conversation cost. Picking a smaller model can keep this near-trivial; the wrong model choice is where teams overspend.
3. WhatsApp conversation charges (the surprise)
Meta charges per conversation window — a 24-hour window opened either by you or by the user. Rates are set per category, per country, and Meta updates them periodically. The categories:
- Marketing conversations — the most expensive tier.
- Utility conversations (order updates, OTPs, reminders) — significantly cheaper than marketing.
- Authentication conversations — comparable to utility.
- Service / user-initiated conversations — a free monthly allowance per WhatsApp Business Account, then a low per-conversation rate.
For current rates, check Meta’s WhatsApp pricing page before quoting any campaign — they vary by country and change periodically. The build cost is amortised over years; the conversation cost is forever.
How to think about total cost of ownership
The honest framing for a buyer in India in 2026:
- If you do low monthly inbound volume: a no-code platform on a small monthly plan is almost always the right answer. Custom build is overkill.
- If you do moderate inbound volume: custom or no-code both work. Custom wins if you have non-standard integrations or care about owning the system.
- If you do high inbound volume or run marketing campaigns: custom usually wins on TCO over a year or two — both because you escape per-seat platform fees, and because you can optimise conversation routing (utility templates instead of marketing where possible).
Where teams overspend
- Sending marketing as marketing when it could be utility. Order updates, appointment reminders, and shipping confirmations qualify as utility — significantly cheaper than marketing per conversation.
- Using a top-tier LLM for intent classification. A smaller, cheaper model does the same job at a fraction of the cost.
- Paying for a fancy BSP dashboard nobody opens. Going direct through Meta Cloud API is realistic for most teams once you have an engineer involved.
What to do next
Before quoting any vendor or platform: estimate your monthly conversation volume and split it by category (marketing vs utility vs service). That single spreadsheet drives every other decision — build vs no-code, BSP vs direct, and which model to use for replies.
One often-forgotten line item: hosting. A small VPS for a Node.js webhook receiver is inexpensive if you pick the right tier. We wrote a separate guide on where to host a WhatsApp bot — what to look for, what to skip, and the setup we use ourselves.
Want a written quote for your volume and use-case? See the WhatsApp chatbot service page for what’s included, or send us your numbers and we’ll come back inside 24 hours.