Restaurants

Restaurant missed calls: stop losing bookings

Restaurant missed calls mean lost phone bookings during service and after closing. How an AI phone assistant takes every request without pulling staff away.

Foxy meditating calmly while the restaurant is closed and the phone is still answered

Restaurant missed calls are rarely about carelessness. The phone rings when every table is full, when the manager is settling a bill for eight, or after closing, when someone on their sofa decides where to eat on Saturday. Many of those calls are phone bookings that end up at the place next door. This guide looks at when they get lost, what they might be worth, and how an AI phone assistant can take every request without anyone putting down a tray.

Restaurant missed calls: when phone bookings get lost

Look at your phone's call log and you'll probably see the same pattern:

  • Mid-service. Spanish service hours, roughly 13:30–15:30 and 20:30–23:00, are exactly when the people who could answer are serving tables, and exactly when most people call.
  • When you're closed. Your day off, and late at night after closing. Plenty of people plan tomorrow's dinner at 11pm.
  • When the line is busy. A long call about a group menu blocks the line while two more callers give up.

We're not quoting a missed-call percentage here: we haven't found an independent, reliable study for the restaurant trade in Spain. What you can do is measure yours. Count the unanswered calls in your log for two weeks. Five minutes of work, and more useful than any industry average.

What an AI phone assistant does in a restaurant

CallSmarter has a restaurant template. This is what the agent does when it answers on your behalf:

  1. Introduces itself as an AI assistant and says the call is recorded. The platform always adds this; it doesn't depend on your script.
  2. Answers questions from your knowledge base: kitchen and dining-room hours, address and parking, menu and set lunch, prices it may quote, terrace, dogs, high chairs, takeaway.
  3. Takes the booking request: name, party size, date and time, a contact number, any allergies or intolerances and any special occasion. It reads the details back before saving.
  4. Leaves the booking pending. It never says the table is booked or that there's availability. It says the restaurant will review and confirm.
  5. Categorises the call: booking, takeaway order, event or group, enquiry, complaint, or other (suppliers, CVs, sales calls).
  6. Alerts the right person via Telegram, email, a CRM task or a webhook.

Allergens: the assistant notes, the restaurant confirms

No shortcuts here. Spanish rules allow restaurants to give allergen information verbally, provided staff can supply it and it is recorded in writing on the premises (Royal Decree 126/2015, Art. 5). So the template is set up so that the assistant never guarantees a dish is suitable or free from traces, even if the menu says so. It notes the allergy in the caller's own words and says the team will confirm. If someone describes an allergic reaction happening now, it tells them to hang up and call 112.

Five example scenarios

Illustrative situations, not customer cases.

Saturday, 21:15, full house. A caller wants a table for six at Sunday lunch. The assistant takes name, party size, date and time, and asks about allergies: one guest is coeliac. The request lands in the team's Telegram group tagged "booking" with the coeliac note. At 23:30 the manager reviews requests and confirms by message.

Monday, closed. Three callers ask whether you open on Tuesday. The assistant answers from your hours. Nothing to action, so they're logged as "enquiry" with no alerts. Nobody's phone buzzes.

Takeaway during service. Someone wants to collect two pizzas at 20:45. In the template, a takeaway order while you're open triggers an instant alert, because the kitchen needs to know now.

Company lunch for 25. The assistant takes the basics and the date and tags it "event or group". Your rule can assign it to the events manager and create a follow-up task for the next morning.

A complaint. An unhappy customer calls about an order. The assistant doesn't argue or promise refunds: it listens, records the reason and flags it high urgency. Complaints are alerted immediately.

What an unanswered booking is worth

Run the numbers with your own data. The "Inbound value" report in the dashboard estimates:

Saved revenue ≈ booking requests × average ticket × close rate

Worked example (assumed figures): a restaurant receives 10 booking requests a week outside service hours. Average party of 3 at €30 a head (€90 per booking), and 8 in 10 confirmed (close rate 0.8): 10 × 90 × 0.8 = €720 a week. If your figures are half that, so is the estimate. The point is to use your numbers. The template ships with suggested close rates per category (0.8 for bookings, 0.4 for group events) that you adjust. For the full method, see the cost of a missed call.

Setting it up without losing a service

  1. Write the knowledge base by section: hours, address, menu, prices, booking policy (maximum group size, sittings, deposits, cancellations, grace period), allergens, services and house rules. One afternoon is enough.
  2. Decide who gets what. A sensible start: bookings and events to the floor team's Telegram group; takeaway during service, instant alert; complaints to the manager; other messages to email.
  3. Set category values (average ticket and close rate) so the report makes sense.
  4. Test it with your staff. Let them try to catch it out: groups of fifteen, unusual allergies, questions that aren't on the menu.
  5. Add it to your website. The live channel today is the "Call the AI" button on your site, usable from a phone or laptop with nothing to install. Answering directly on your phone number, with forwarding when you don't pick up, is being rolled out; ask us about availability.

What it doesn't do yet

  • It doesn't check or block tables in your reservation system; your team confirms.
  • It doesn't transfer the call live to a person; when needed, it flags the request as urgent and promises a call back.
  • It doesn't take deposits or card details.

If your restaurant lives on confirming tables within seconds, the best use today may be covering the hours when nobody answers, not replacing the floor phone.

Tone matters as much as tech

People choose restaurants partly on how they're treated on the phone. The template uses short sentences, one question at a time and full dates ("Saturday 14 November at 21:00") to avoid mix-ups. Every conversation is recorded, transcribed and scored, so you can listen to how your guests were treated and change what you don't like.

Want to hear it first? Talk to Foxy on our site. For the bigger picture across industries, start with our guide to the AI receptionist.

Frequently asked questions

Does the assistant confirm the table?

No. It records the request with all the details and tells the caller that the restaurant will review and confirm it. You never promise a table you don't have.

What if a caller asks about allergens?

The assistant never guarantees a dish is safe. It notes the allergy or intolerance in the caller's own words and explains that the staff will confirm. Allergen information remains the restaurant's responsibility.

Can it answer tourists in English?

Yes. It replies in the caller's language; we support Spanish, English, Russian, Catalan, Portuguese, French, German, Italian and Hebrew.

How do I find out about a new booking request?

Through the channel you choose: a team Telegram group, email, a CRM task or a webhook into your system. Takeaway orders during service can trigger an instant alert.

Sources

  1. Royal Decree 126/2015 on food information for non-prepacked foods (BOE)
  2. European Commission: transparency obligations under Article 50 of the AI Act