WhatsApp Chatbot for Restaurants (2026)

Because a phone line only works if someone is free to answer it, and during service that is exactly when nobody is. A guest calling at 19:30 on a Friday to change a party size often gets a busy line or a rushed answer between courses being run.

The practical scenario: a party of four with an 8pm reservation messages at 19:45 asking to add two more people and whether the patio is available. A host mid-service cannot always drop what they are doing to answer a call, but a WhatsApp bot checks the floor plan and current bookings instantly and either confirms the change or tells the guest a host will confirm shortly - either way, the guest gets an answer without the restaurant's phone line being tied up.

The same thread carries the reservation confirmation days earlier, so by the time the guest is messaging about the party size change, the conversation already has their booking details attached - no re-explaining who they are or what they booked.

It also fits how a walk-in waitlist already works in most restaurants - a guest gives a name and a number, waits, and is called. Moving that same interaction to WhatsApp means the guest is not stuck standing by the host stand watching for their name to be called; they can wait at the bar next door and get a message the moment the table is free.

Keep the automated scope to bookings and information that follow fixed rules:

• Reservation booking and changes. Show real table availability, confirm in-thread, write the change back to the booking system.
• Confirmation and reminder with a reply path. Confirm, cancel or adjust party size without a phone call.
• Waitlist management. Add a walk-in party to the list and message them when a table is ready - the timing here matters more than almost any other message the bot sends.
• Menu, hours and parking questions. Grounded on your actual menu and hours, not a stale PDF.
• Private event or catering inquiry capture. Collect date, party size and budget range, then route to the events contact.
• Loyalty offers and promotions. Sent to guests who have opted in, tied to your actual loyalty program.

Notice what is missing: confirming a dish is safe for a specific allergy, verifying legal drinking age, and resolving a complaint about food quality. Those need a person on the floor, not a chat thread.

A simple test for what belongs on this list: does the answer come from your booking system or your published menu, or does it require someone to physically check something in the kitchen or make a judgement call about a guest? The first automates well; the second needs a person every time, no matter how routine the question sounds.

This is the rule that decides whether your reminder actually lands before the reservation, and restaurants sending same-day nudges are the ones who feel it most.

When a guest messages you - or opts in when booking - a 24-hour customer-service window opens, during which you can reply freely. Once it closes, only an approved message template can send.

• A reservation confirmation or same-day reminder is utility, tied to a booking the guest already made.
• A waitlist "your table is ready" alert is also utility, but it is uniquely time-sensitive - a template delayed by review queues is useless if the table has already been reassigned, so these need to be pre-approved and tested well before a busy night, not requested during one.
• A loyalty offer or "we miss you" promotion is marketing, business-initiated and promotional, requiring explicit opt-in and stricter review.
• The moment the guest replies to any message, the window reopens and questions about the reservation or menu can be answered freely.

Submit the waitlist and reminder templates first, since they carry the operational weight of a live service night; treat marketing templates as a separate, lower-urgency track.

Frequency discipline matters for the marketing side specifically. A restaurant sending a promotional offer every week to the same guest list will see opt-out rates climb, and once a guest opts out of marketing messages they may stop receiving reminders too if the categories were not separated cleanly from the start - plan that separation before the first campaign goes out, not after complaints start.

The line here is liability and things only staff can physically verify.

• No definitive allergy confirmation. The bot can share listed allergen information from your menu, but it must never tell a guest a dish is "safe" for a specific allergy - cross-contamination and recipe changes are a kitchen judgement, and a wrong answer here is a safety incident, not a bad review.
• No age verification. WhatsApp has no reliable way to verify a guest is of legal drinking age, so alcohol orders or age-restricted promotions must be confirmed in person, never assumed from a chat reply.
• No complaint resolution. A guest unhappy about food quality or service needs a manager who can actually comp a dish or make it right - a bot apology does not fix the meal.
• No overbooking judgement calls. The bot should show real availability only; it should never be allowed to "squeeze in" a booking past capacity to avoid disappointing a guest.

Build the refusal path so an allergy question gets a clear "let me connect you with the kitchen" rather than a confident-sounding guess - a guess is the failure mode that turns a chatbot into a liability.

Health and food safety complaints deserve the same treatment as allergy questions. If a guest reports something that sounds like a food safety issue rather than a simple quality complaint, that should route to a manager immediately and not sit in a general complaint queue behind less urgent messages.

1. Any allergy or dietary safety question. Route to the kitchen or a manager every time, no exceptions.
2. Complaints or negative sentiment. Detected by keyword and by tone - a frustrated guest should reach a person quickly, not loop through menu options.
3. Large parties or private events. Anything beyond standard table booking needs a person who can negotiate space and pricing.
4. Explicit request. "Can I talk to someone" must work at every step.

The handover needs to carry the reservation details and what has already been discussed, so a host is not asking a guest to repeat their party size and time a second time during a busy service. On Conferbot the conversation moves into a shared agent inbox with the full thread attached.

State clearly when nobody is monitoring the thread - between services, for instance - so a guest messaging at 15:30 does not assume nobody read it and call somewhere else instead.

1. Get a WhatsApp Business API number. Separate from any personal phone the host stand uses.
2. Submit templates early. Reservation confirmation, same-day reminder, waitlist-ready alert, and loyalty offer (marketing, opt-in) - request all four before your next busy season, not the week of it.
3. Build the flow once using Conferbot's template library for reservations, waitlist and menu Q&A as a starting structure.
4. Feed reservation and waitlist availability in over webhook so it's real-time - a bot confirming a table that does not exist damages trust faster than a phone line ever could.
5. Write the allergy and complaint escalation copy deliberately - this is the safety-critical text, and it deserves the same care as the booking flow.
6. Pilot during a slower service before relying on it for a Friday or Saturday night.

Because the same flow deploys to every channel on Conferbot, a guest who finds you through your website can book on WhatsApp without re-entering their details - useful for restaurants running both a booking widget and a WhatsApp line, since the two no longer need to be treated as separate booking systems with separate availability.

Track the numbers that reflect covers filled and staff time saved, not just messages sent:

• No-show rate before and after, for reservations that received a confirmed reminder.
• Phone call volume for bookings and simple questions the bot now handles directly.
• Waitlist-to-seated time accuracy - how close the bot's "your table is ready" alert lands to the actual table being free.
• Escalation rate by reason, especially allergy and complaint categories, since those deserve individual attention regardless of volume.

Watch allergy-escalation rate as a safety metric, not a conversion metric - a bot that never escalates an allergy question might mean guests are not asking, or it might mean the detection is missing them, and that difference is worth confirming directly by reading a sample of transcripts rather than trusting the number alone.

Can a WhatsApp bot confirm if a dish is safe for an allergy?

No. It can share the allergen information listed on your menu, but it should never tell a guest a dish is definitively safe, because cross-contamination risk and recipe changes are a kitchen judgement call. Route any allergy question to a manager or the kitchen directly - a confident but wrong answer here is a safety incident, not a bad review.

What is the 24-hour window for restaurant WhatsApp messaging?

It is the period after a guest messages you, or opts in when booking, during which you can reply with anything. Once it closes, only pre-approved message templates can send. Reservation reminders and waitlist alerts are utility templates tied to an existing booking; loyalty offers are marketing templates requiring opt-in and stricter review.

Can a chatbot manage a restaurant waitlist?

Yes, and it is one of the strongest use cases - the bot adds walk-ins to the list and sends a 'your table is ready' alert the moment a table frees up. Because that alert is uniquely time-sensitive, get the message template pre-approved well before a busy night; a template stuck in review is useless once the table has already been reassigned.

Should a restaurant bot handle complaints about food?

No, not the resolution. The bot should recognize a complaint by keyword or tone and route it to a manager immediately, since only a person can actually comp a dish, remake an order, or otherwise make the situation right. A bot apology on its own tends to frustrate a guest further rather than resolve anything.

Can a WhatsApp bot take alcohol orders?

It should not process age-restricted orders on its own, because there is no reliable way to verify legal drinking age over a chat thread. The bot can note the request and flag it for staff to confirm in person, but age verification itself needs to happen face to face, not through an automated conversation.

How much does a WhatsApp chatbot cost for a restaurant?

There are two costs: the chatbot platform and WhatsApp's own conversation fees, billed by Meta per category and destination country. Plans are $19, $39 and $59 a month, and the free tier covers 600 conversations on the website widget. WhatsApp needs the Business plan - the pricing page lists what each tier includes. WhatsApp API messaging fees are billed by Meta directly.