Restaurants & Food Service

1. Overview

In the restaurant and food service industry, voice AI is primarily used to automate the order-taking process, both over the phone and in drive-thrus. This vertical is unique due to its demand for high speed and accuracy in often noisy environments. A successful voice AI can significantly reduce wait times, improve order accuracy, and increase average order value through automated, consistent upselling.

Phone and delivery orders capture customer names, phone numbers, delivery addresses, and card details. On the Growth plan you can turn on PII redaction to keep them out of stored transcripts.

2. Common Call Intents

This table outlines the most common reasons customers call a restaurant.

Intent Name	Description	Example Caller Utterance
Place an Order	The customer wants to place a new order for pickup or delivery.	”I want to order a large pepperoni pizza and a side of wings.”
Check Order Status	The customer is calling to check on the status of an existing order.	”I placed an order about 20 minutes ago and I was just wondering if it is ready for pickup.”
Make a Reservation	The customer wants to book a table at a sit-down restaurant.	”I would like to make a reservation for two people at 7 PM tonight.”
Ask About a Menu Item	The customer has a question about a specific menu item, such as ingredients, price, or allergens.	”Does your cheeseburger have onions on it? And is it gluten-free?”
Get Store Hours or Location	The customer wants to know the restaurant’s hours of operation or address.	”What time do you close tonight?“

3. Common Call Outcomes

This table lists the typical dispositions or final results of restaurant calls.

Outcome Name	Description	Success/Failure
Order Placed	A new order was successfully placed and confirmed with the customer.	Success
Reservation Made	A new reservation was successfully booked and confirmed.	Success
Information Provided	The customer’s question was answered, and the issue was resolved.	Success
Transferred to Staff	The call was transferred to a human employee for further assistance (e.g., for a complex issue).	Neutral/Failure
Caller Hung Up	The caller disconnected before their order or inquiry was completed.	Failure
Order Not Placed	The caller decided not to proceed with placing an order.	Failure

4. Recommended Evaluation Criteria

This section breaks down key evaluation criteria into specific, measurable checks for monitoring and improving agent performance.

Start from the eval library. Part of what follows is already covered by Tuner’s predefined evals, which you enable with one click: Data capture accuracy, Confirmation accuracy, Hallucination, Tool args mismatch, Unnecessary repetition, and Conversation flow and turn taking. Between them they cover the items and modifications as captured, the read-back at the end, invented menu or price details, what was sent to your ordering system, and how many turns the order took.Enable those first, then add the rest as custom evals. The criteria in the tables below are written to be used as the prompt definition of a custom eval.

Order & Transactional Accuracy

Evaluation Name	Description	Type
Order Item Accuracy	Verifies that the agent correctly identified and added every food and drink item the customer requested to the order without omissions or incorrect items. This is fundamental to order accuracy and customer satisfaction.	Pass/Fail
Order Modification Accuracy	Evaluates whether the agent accurately captured all special requests and modifications such as “no onions,” “extra cheese,” “sauce on the side,” or other customizations. Accurate modification capture is critical for customer satisfaction.	Pass/Fail
Price Calculation Accuracy	Confirms that the agent quoted the correct total price for the order, including all items, modifications, applicable taxes, and any additional fees. Price accuracy builds customer trust and prevents billing disputes.	Pass/Fail

Conversational Quality & Efficiency

Evaluation Name	Description	Type
Upsell Effectiveness	Assesses the agent’s ability to make relevant and contextually appropriate upsell attempts, such as suggesting a drink with a combo meal or dessert with an entree. Higher scores indicate better sales techniques and successful upsells that enhance revenue.	Scored 1-5
Ordering Efficiency	Evaluates whether every turn moved the order forward. Penalize re-asking for details the customer already gave, repeating the full order more often than needed, and detours that add turns without adding information. Wasted turns are the largest controllable driver of handle time on high-volume phone and drive-thru orders.	Scored 1-5
Handling of Ambiguity	Measures the agent’s ability to recognize and clarify unclear customer requests by asking appropriate clarifying questions (such as “Did you say large or medium?”) rather than making assumptions that could lead to order errors.	Pass/Fail
Misrecognition Recovery	Evaluates whether the agent caught and repaired a mis-heard item instead of carrying it into the order. Look for the customer correcting the agent (“no, I said large”) or a read-back that doesn’t match what was asked for, and whether the agent re-confirmed before moving on. Phone and drive-thru orders are noisy, and recovery is what separates a corrected order from a wrong one.	Pass/Fail

5. Compliance & Regulatory Requirements

Payment Card Industry Data Security Standard (PCI DSS): For any phone orders involving credit card payments, the voice AI and payment processing system must be PCI DSS compliant to ensure the secure handling of cardholder information.

6. Key Performance Benchmarks

Order Accuracy: 95% or higher
Upsell Acceptance Rate: 40-50%
Average Handle Time (AHT): Under 2 minutes for typical phone orders

Getting Started

Integrations

Evaluation

Observability

Simulation

MCP

Guides

Resources

1. Overview

2. Common Call Intents

3. Common Call Outcomes

4. Recommended Evaluation Criteria

Order & Transactional Accuracy

Conversational Quality & Efficiency

5. Compliance & Regulatory Requirements

6. Key Performance Benchmarks

​1. Overview

​2. Common Call Intents

​3. Common Call Outcomes

​4. Recommended Evaluation Criteria

​Order & Transactional Accuracy

​Conversational Quality & Efficiency

​5. Compliance & Regulatory Requirements

​6. Key Performance Benchmarks

1. Overview

2. Common Call Intents

3. Common Call Outcomes

4. Recommended Evaluation Criteria

Order & Transactional Accuracy

Conversational Quality & Efficiency

5. Compliance & Regulatory Requirements

6. Key Performance Benchmarks