Key takeaways ✺
- A full-time human receptionist costs about ₹15,000 to ₹25,000 a month for roughly 8 hours a day. Covering every hour of the month takes around 3.5 people.
- An AI receptionist has two costs: Meta's message fees, which are small, and a software fee that varies, so get a written quote.
- Faster replies convert better, but the famous multipliers aren't reliable. Measure your own numbers.
- Use AI for first replies and bookings, and hand difficult chats to a person.
Every few weeks someone asks us the same thing: should we hire a person to handle WhatsApp, or put an AI on it? We went looking for hard numbers to answer it properly. Some of what we found is solid. A lot of what gets repeated online isn't, and we'll say which is which.
What a human receptionist really costs
In India, a full-time receptionist is commonly quoted at ₹15,000 to ₹25,000 a month. That is one person, one shift, about 8 hours a day, six days a week. It varies by city, and we found this range in one Indian source, so treat it as a ballpark.
Do the sum. ₹20,000 over about 208 working hours a month is roughly ₹96 for every hour covered. A month has 720 hours. To cover all of them you'd need about 3.5 people, which is around ₹69,000 a month in salary alone. On top of that come PF and other statutory contributions, paid leave, and the cost of finding someone new when they leave. Those are real, but they depend on your setup, so we won't put a made-up number on them.
What an AI receptionist costs
There are two parts, and only one of them is easy to pin down.
Meta's message fees. Since 1 October, service replies are free up to 1,000 a month per business number, then ₹0.1150 each plus 18% GST. A business sending 3,000 replies pays for 2,000 of them: about ₹230, or roughly ₹271 with GST. Marketing messages cost more, at ₹0.8631 each.
The software fee. This is the part that varies. The only public prices we could find were American, from about $99 to $899 a month depending on features. Indian providers mostly quote on request. So ask for a written quote, and ask what happens when you go over the limit.
Does it convert better? Be careful here
This is where most articles get loose, so here is what we can and can't say.
- Speed matters. Almost every source agrees that the faster you reply, the more enquiries turn into customers, and that most of the drop-off happens in the first hour.
- The exact multipliers aren't reliable. The most quoted one is "21 times more likely to qualify a lead if you reply within 5 minutes". It traces back to an old InsideSales and MIT study, and we couldn't find the original. Other blogs claim 8 times, 100 times, even 400% for the same idea.
- WhatsApp conversion figures are all over the place. We saw 7 to 12%, 8 to 15% and 28%. Most come from companies that sell chatbots. We found no independent study that isolates what an AI does to conversion on WhatsApp.
- After-hours enquiries are real, but yours may differ. One vendor case study, a resort, reported about a third of enquiries arriving outside working hours. That is one business and the vendor's own data.
So here is what we'll stand behind: an AI that replies in seconds, at 2 am, will almost certainly catch enquiries that a person on an 8-hour shift misses. How many extra customers that means for you, nobody can tell you from the outside. You have to measure it.
Where a person still wins
An AI is weaker with upset customers, unusual requests and anything that needs judgement or a bit of warmth. It can't greet someone who walks in. That is why most good setups use both: the AI takes the first reply, answers the usual questions and books appointments, and hands the chat to a person when it gets difficult.
Why it is still worth it for most businesses
- It replies in seconds, at any hour, including Sundays and festival days.
- It gives the same answer every time and doesn't forget to follow up.
- It doesn't get overwhelmed. Five chats or five hundred, the wait is the same.
- It frees your staff from answering "what are your timings?" forty times a day.
- The cost grows with use, not with a fixed salary you pay even on a quiet month.
A simple way to find out
Pick one month and write down three things: how fast you reply now, how many enquiries arrive after closing time, and how many turn into bookings. Run the AI for the next month and compare. It takes a spreadsheet and ten minutes a week, and your own numbers will tell you more than any statistic in this post.
Where the figures came from: salaries from MNB Research's guide for Indian SMEs, US software pricing from several vendor guides, WhatsApp conversion claims from Aurora Inbox and SetSmart, and Meta's fees from its pricing page.
Questions people also ask
Can an AI receptionist fully replace a human one?
Not for everything. It is weaker with upset customers, unusual requests and walk-ins. Most businesses use it for first replies, common questions and bookings, and pass difficult chats to a person.
How much does Meta charge for WhatsApp replies in India?
Since 1 October, the first 1,000 service replies a month per business number are free. After that each costs ₹0.1150 plus 18% GST. Marketing messages cost ₹0.8631 each.
Is there proof that an AI improves conversion on WhatsApp?
Not independent proof. The figures we found come mostly from chatbot vendors and range from about 7% to 28%. Faster replies do help, so track your reply time and bookings for a month before and after.
How do I find out how many enquiries I miss after hours?
Open last month's chats and count the first messages that arrived after closing time, then check how many of those got a reply the same day.