WhatsApp for restaurants and cloud kitchens: orders, bookings and the review you never saw coming

Eight o’clock on a Friday at a 40-seat restaurant in Indiranagar that also runs a cloud kitchen brand from the same stove. The kitchen phone has three missed calls. A WhatsApp message on the owner’s number says “2 butter chicken 4 naan Koramangala 5th block pls fast”. An Instagram comment under last week’s reel asks “do you deliver to HSR?”. A table of six for 8:30 was booked by message two days ago and nobody wrote it in the book.

The person who knows the answers to all of this is the owner, and she is at the pass.

Restaurants do not have a messaging problem. They have a phones problem: orders, bookings, questions and complaints arrive on three devices held by people whose hands are busy. This playbook moves them to one number, answers the repeat questions automatically, and gives unhappy customers a quiet place to say so before they say it in public.

Which messages actually arrive

Message	Where it lands today	What it should be
”Can I order 2 butter chicken…”	Owner’s WhatsApp, kitchen phone	A direct order, confirmed in one message
”Is my order ready?”	Three calls to the kitchen	Answered by the same message that confirmed it
”Table for 6 at 8:30?”	WhatsApp, Instagram DM, a call	A booking with a reminder on the day
”Do you deliver to HSR?”	Instagram comment, unanswered	A public one-line reply, with the menu sent privately
”The biryani was cold”	A 2-star review, two days later	A private message, an hour after delivery, handed to the manager

The first four are the same question asked in different places. The fifth is the one that costs real money, and it is the only one most restaurants never receive directly.

Step 1: one business number, several hands

Put one number on the WhatsApp Business Platform, Meta’s service for businesses, through a provider. Unlike the free WhatsApp Business app, several people answer from a shared inbox: the cashier during service, the owner on the phone between tables, the night manager for the cloud kitchen. Each chat is assigned to one of them, so two people never confirm the same order.

Use the number that is already on Swiggy, Zomato and Google Maps as the “restaurant phone”. Add it to the menu card, the bill, the delivery bag sticker and the Instagram bio. The point is not to replace the aggregators overnight. It is that a customer who has ordered from you twice through an app should have a way to order a third time without the commission.

Step 2: direct orders, confirmed in one message

A direct order is only worth taking if the customer trusts it more than the app. That trust is built by speed. The flow:

1. The customer messages the number. A list message offers the menu by section, or a single link opens the full menu on your website (a chat widget there lands in the same inbox).
2. The order arrives as a chat. Whoever owns the inbox during service taps one reply: “Got it: 2 butter chicken, 4 garlic naan. Total ₹860, delivery in about 35 minutes. [Confirm order] [Change order]”.
3. The customer taps Confirm. The chat is tagged with the order and the area, and a note goes to the kitchen board, which is whatever the kitchen already uses, from a printed ticket to a tablet.
4. Payment is the UPI or payment link you already use, sent in the same chat.

Then the message that removes the phone calls: when the rider leaves, a journey sends “On its way, Vikram. Your rider is Imran.” That is the answer to “is it ready?” before the question is asked.

Confirmations and “out for delivery” are utility messages, approved once as templates by Meta and charged at ₹0.1150 each on the India rate card. The customer’s own messages are service messages, free up to 1,000 per number each month.

Step 3: table bookings with a reminder on the day

A table request is a booking in two taps:

Hi Riya, for 6 people tonight we have: [8:00 pm] [8:45 pm] [9:30 pm]

Confirm it in the inbox, write it in the book, and let a journey send the reminder at 4 pm: “See you at 8:45 pm at Masala Bay, table for 6. [Confirm] [Cancel]”. A Cancel at 4 pm puts the table back on sale for the dinner service. A no-show at 8:45 does not.

Step 4: the feedback message that replaces the 2-star review

Most unhappy customers never tell the restaurant. They tell the app, in public, two days later. Give them a private route first.

Sixty minutes after delivery, or at closing time for dine-in, a journey sends:

How was tonight’s order from Masala Bay? [Loved it] [It was fine] [Something went wrong]

Then branch on the answer:

• Loved it gets a thank-you and, if they opted in, a nudge towards the Google review page.
• It was fine gets nothing. Do not over-message.
• Something went wrong hands the chat to the manager, with the order beside it. The manager replies within the hour, from a person, with a fix: a replacement, a credit on the next order, or a straight apology. Most people who are heard do not then write the review.

Keep the wording factual and the buttons short, and Meta approves it as a utility template because it is about a transaction the customer made.

Step 5: Instagram comments are orders and bookings

Food reels get comments, and most of them are questions with money behind them: “do you deliver to HSR?”, “price of the thali?”, “open on Diwali?”. On a platform that supports Instagram comments, each one arrives in the inbox as a conversation with the reel attached.

Answer publicly in one line, so everyone scrolling sees that you reply: “Yes, we deliver to HSR till 11 pm. Sent you the menu.” Then send the menu link privately, as one Direct message within Meta’s 7-day window. If the comment is spam, hide it; if it is abuse, delete it after the confirmation prompt. All of this is done by the same person who is answering WhatsApp, from the same screen.

Step 6: repeat customers, with permission

A customer who has ordered three times directly is worth a message on a slow Tuesday. That message is a marketing message: it needs an opt-in, an approved template and the marketing rate of ₹0.8631 each. Ask for the opt-in in the thank-you after the first direct order, record it on the contact, and send offers to a segment such as “ordered twice or more, opted in, not ordered in 30 days”. Fifty people, not the whole list. The opt-in guide has the wording.

What it costs for a month

A restaurant doing 600 direct orders a month plus 150 table bookings:

Message	Category	Volume	Meta cost
Customers’ messages and orders	Service	~1,500	₹0 for the first 1,000, then ~₹58
Confirmations, out-for-delivery, booking reminders	Utility	~1,500	~₹173
Feedback messages	Utility	~600	~₹69
Tuesday offer to opted-in regulars	Marketing	~200	~₹173

About ₹470 a month in Meta charges before GST, plus the provider’s plan. Against aggregator commission on 600 orders, the arithmetic is not close.

How this runs in Conecxt

The pieces above are live features, not plans: a shared inbox for WhatsApp, Instagram comments and Direct messages, and a chat widget on your website menu, with one record per customer; list messages, reply buttons and approved templates; journeys that send the confirmation, the rider update, the booking reminder and the feedback message on schedule and hand the bad answers to the manager; and segments for the regulars’ offer. An AI agent can take the first turn on “do you deliver to…” and the menu, and stays off until you switch it on.

Start with Steps 2 and 4. Direct orders pay for the plan; the feedback message pays for the reputation.