Sending a message to 5,000 people takes minutes. The hard part is what comes next: who read it, who replied with a question at 9 pm, and who quietly dropped off after week two. That’s the gap WhatsApp chatbots fill. About 150 NGOs on Glific use them today, and together they have sent over 183 million messages. Here’s what they’ve learned.
What a WhatsApp chatbot actually does
It’s an automated conversation on your organisation’s WhatsApp number. It asks questions and saves the answers, sends each person content that fits them, collects photos and voice notes, and follows up on its own, days later. Some chatbots follow menus your team writes. Others use AI to understand questions in people’s own words. Many NGOs use both.
What NGOs use them for
More than half the NGOs on Glific work in education, sending activities to parents and on-demand help to teachers. Others guide frontline health and child-protection workers, answer farmers’ questions, replace paper forms with WhatsApp check-ins, and let families report damage after a cyclone. The full guide maps ten program areas to real examples.
The cost question most NGOs get wrong
There are two costs: a platform fee (on Glific, ₹15,000 to set up, then ₹9,500 a month or ₹1,04,500 a year) and Meta’s fee for each message you send. Since 1 October 2026, chatbot replies are charged too, after the first 1,000 a month on each number. What surprises most teams is how much one setting changes the bill.
The full guide walks through this example line by line, shows what a first year costs, and explains how to get your message categories right.
From idea to live chatbot in about 3–4 weeks
Most of the time goes into planning, not technology: one clear goal, one person who owns the chatbot, and a conversation tested by hand on WhatsApp before anything is built. Meta’s business verification takes 3–4 working days, and the most common reason it gets rejected is easy to avoid once you know it.
What 16 NGOs learned piloting AI
In one pilot, a manual audit found 91.8% of AI answers to facilitators’ career questions accurate. In another, farmers in Sikkim and Darjeeling asked their questions by voice note in Nepali. AI also needed work: urgent messages needed fixing, tested replies, and answers stayed grounded in the right material only after clearer prompts and repeated testing. Getting AI right is an iterative process, not a one-time setup.
Why chatbots stall
Rarely because of the technology. The most common reason is that no one owns the chatbot, so it slowly goes out of date. The full guide covers six common mistakes and what to do instead.
Get the full guide
Download the complete guide for the worked cost example and first-year budget, a Meta verification checklist, four NGO case studies (including what didn’t go to plan), and a checklist to see whether a chatbot fits your program:
 | Thank you for Signing Up |
Or book a 30-minute call to estimate costs for your own program.