Sample result
WhatsApp Bot Implanter for Restaurants
The owner asked this:«About 60 messages a day come in across Instagram and WhatsApp. We answer when we can; at the lunch peak nearly all of them drop. I know we're losing delivery orders but I don't know how many.»
① The verdict
| Messages a day | 60 |
| Unanswered at the peak (12:00–14:30) | ~26 |
| Of those, asking a price or whether you deliver | 19 |
| Average delivery ticket | $38,000 |
| If 1 in 4 is recovered | ~$5.7M/month |
| Bot cost (machine + channel + model) | ~$310,000/month |
It pays for itself in the first week, and not because of the bot: because of the nineteen questions nobody answers today that are answered by READING a value. Caveat: the figure above comes from your 60 messages and my peak count — if the real count is half, it still pays.
② The question inventory (extract)
| Question | Channel | Times/wk | How it gets answered | Data needed |
|---|---|---|---|---|
| «how much is the platter?» | delivery | 84 | NO MODEL | menu with prices |
| «do you deliver to Laureles?» | delivery | 61 | NO MODEL | zones with cost and time |
| «how late are you open?» | direct | 38 | NO MODEL | hours with the Sunday exception |
| «where are you?» | direct | 22 | NO MODEL | address and landmark |
| «anything without pork?» | delivery | 14 | MODEL | menu with descriptions |
| «can you replace an order that arrived cold?» | delivery | 6 | TO A PERSON | — |
The cut: 4 out of 6 need no model. Those cannot invent a price because they do not think: they read. And the first two rows are 65 % of the week's volume: that is the 20 % that must be perfect before switching on.
The full example has 1 more part(s): you see them inside the library, with your account.