Restaurants and Cafes: Handling Reservations, Menus and Delivery Questions in Your DMs

Ask anyone who manages a restaurant's Instagram what the DMs look like on a Friday evening. "Table for 6 at 8?" "Do you deliver to Jhamsikhel?" "Is the momo platter still Rs 450?" "Are you open on Tihar?" All while the floor is full and the one person who knows the answers is carrying plates.

Restaurant messages are a special kind of busy. They are short, repetitive and time-sensitive. A diner asking about a table tonight won't wait three hours. This guide covers how to handle them without hiring someone just to sit on the phone.

What people actually message restaurants about

Look through a month of your DMs and you will find most messages fall into a handful of groups:

• Reservations. Table for how many, what time, any space outside, can we bring a cake.
• Menu and prices. Do you have veg options, what's in the set lunch, is there a kids' menu.
• Hours and location. Open today, closing time, parking, how to find you.
• Delivery and takeaway. Do you deliver here, minimum order, how long, which apps.
• Events and groups. Birthday packages, office lunches, private bookings.
• Feedback. Compliments, and the occasional complaint about a late order or a cold dish.

The first four are almost entirely routine. The last two need a person, at least for part of the conversation. That split is the basis for everything below.

Build a menu and info sheet the bot can use

A chatbot can only be as good as the information it has. For a restaurant, that means writing down, in one place:

1. The current menu with prices, and which items are veg, vegan or contain common allergens, as far as you know them.
2. Opening hours, including holiday closures and changes during festivals.
3. Address, a landmark-based direction ("next to the big temple, second floor"), and parking notes.
4. Delivery areas, delivery charges, minimum order and rough delivery times.
5. Which delivery apps you are on, if any, and whether direct orders are possible.
6. Reservation rules: largest table, whether you hold tables for late guests, deposit for big groups.

Keep it updated. The biggest source of angry restaurant DMs is a price or dish the bot quoted that changed last month. Our knowledge base guide has a longer checklist.

Handling reservations in chat

There are two sensible ways to do this.

Collect and confirm

The bot asks for name, phone number, date, time and number of guests, then tells the customer a staff member will confirm shortly. Your host sees the request in the inbox and replies "Confirmed, see you at 8". This works for most small restaurants and cafes, because the person managing tables stays in control.

Instant confirmation

If you use a booking system with live availability, the bot can check and confirm on the spot. It's smoother, but only worth it if your table data is reliable. A bot confirming a table you don't have is worse than no bot.

Either way, keep the questions short. Ask for everything in one message rather than five separate ones:

Happy to book that for you! Please send your name, phone number, date, time and how many people. We'll confirm within a few minutes.

Delivery questions without the back-and-forth

Delivery is where customers get impatient. They want to know three things quickly: do you come to my area, how much, how long. A good bot answers all three in one reply, then offers to take the order or points to the app.

Say a customer in Lalitpur writes "Bhaktapur ma delivery huncha?" The bot should understand that Romanized Nepali question, check your delivery zones, and answer honestly, even if the answer is no. "Sorry, we only deliver inside Kathmandu and Lalitpur right now, but takeaway is ready in about 20 minutes" is a perfectly good reply. More on mixed-language messages in our Romanized Nepali guide.

For an international example, picture a brunch cafe in Sydney getting "do you have gluten free options and can I book for Sunday 10am?" in a single message. The bot should handle both parts, not answer one and ignore the other.

Comments on food posts

Restaurant posts get a lot of comments like "price?" and "where is this?". Answering each one publicly with the full menu clutters the post. A cleaner approach is to reply briefly in public and send the details by DM. That is exactly what comment-to-DM does. It works especially well for new dish launches and festival specials, when one post can pull in dozens of the same question.

What to keep human

• Complaints about an order. A cold dish or a late delivery needs an apology from a person, and often a refund or a voucher. See our guide on handling complaints in DMs.
• Serious allergy or dietary questions. The bot can share what you've written, but should always suggest the guest confirm with staff. Never let it promise something is safe.
• Large groups and events. Office parties and wedding dinners involve negotiation and custom menus.
• Anyone who asks for a person.

Mistakes restaurants make with automation

• Sending the full menu as a wall of text. Link to a menu image or page and answer the specific question.
• Forgetting holiday hours. Update the bot before Dashain, Tihar and public holidays, not after the first confused customer.
• Confirming tables nobody checked. If staff confirm reservations, the bot must say "we'll confirm shortly", not "you're booked".
• Splitting channels. Reservations on WhatsApp, delivery questions on Instagram, complaints on Messenger. Put them in one unified inbox so the host sees everything.

If you run a restaurant or cafe and the DMs are getting away from you, have a look at how Orbit works for restaurants and food businesses. Our team sets it up with your menu and rules, and you can book a demo to see it answer questions from your own page.