Food service and hospitality: full guide

How each claim is supported:

EstablishedPublished resultSurveyVendor claimIllustration

Tap any marked phrase for its source.

Section 1 of 9

How this industry makes money

In one sentence

Hospitality earns on traffic times the average check, and spends most of it on food and labor. AI touches forecasting, scheduling, ordering and marketing; it touches guests only with care.

Sales equal guests times what each guest spends. Costs come mostly from food, labor and rent. Most operators keep a small share as profit, so the biggest levers are the biggest costs.

Six pools of money decide the year.

Pool	What it is	Who controls it
Traffic	How many guests come, and how often they return	Guests, shaped by service, value and marketing
Average check	What each guest spends	The menu, the staff and the price
Food and drink cost	What goes on the plate or in the glass, plus waste	The kitchen, the bar and suppliers
Labor	Wages, benefits and scheduling	Managers, within local wage rules
Occupancy	Rent, energy and upkeep	Landlords and utilities
Fees	Card fees and delivery app commissions	Payment firms and delivery apps

More than nine in ten operators name food, labor, insurance, energy and card fees as significant challenges.

AI touches three of these most: labor, through scheduling; food cost, through ordering and waste; and traffic, through marketing. It touches the average check and the guest's experience too, but that is where it has gone wrong most publicly.

Next: Where AI touches the economics ›

Section 2 of 9

Where AI touches the economics

In one sentence

AI helps most with scheduling, ordering and marketing, where guests do not see it. Where guests do see it, in ordering and pricing, it can cost more in trust than it saves.

Labor: schedules that follow traffic. AI tools predict how busy each hour will be and suggest shifts to match. Most operators using AI say it cut labor costs. The risk is cutting too close, so service slips on a busy night.

Food cost: less waste. AI can forecast how much to prep and order, based on past sales, weather and events. Most AI users report lower food costs.

Traffic: cheaper marketing. Marketing is the most common use of AI in restaurants. Posts, emails and offers cost less to make, so small operators can market more often.

Ordering: guests notice mistakes. Only 6% of operators use AI to take orders. Large chains that tried voice AI at the drive-thru have pulled back or limited it.

Price: guests punish surprises. Wendy's faced a backlash in 2024 for suggesting prices that change by demand. Hotels have long priced rooms by demand, and guests accept it there. A bar or restaurant that tries the same needs to explain it, and should only lower prices at quiet times.

Time saved is not money saved. Hours freed for a manager become money only if shifts change or the manager uses the time to sell more.

Next: What is already happening ›

Section 3 of 9

What is already happening

In one sentence

The back-office uses are spreading and operators report savings. The guest-facing ones have been pulled back or limited when guests did not like them.

Who	What they did	Reported result	What the source does not show
US restaurant operators (National Restaurant Association, 2026)	Adopted AI tools	26% use AI; marketing is the top use; 6% use it for orders	What it saved
Operators using AI (Restaurant365, 2026)	Used AI for scheduling, inventory and reporting	62% report lower labor costs; 61% lower food costs	How much lower, and a comparison with operators not using it
McDonald's (2024)	Tested voice AI at the drive-thru in more than 100 restaurants	Ended the test while saying voice ordering is still in its future	Accuracy and cost figures
Taco Bell (2025)	Put voice AI in more than 500 drive-thrus	Said it is reconsidering where to use it, and that people may do better at busy locations	Sales or speed figures
Wendy's (2024)	Said it would test dynamic pricing on digital menu boards	Public backlash; it promised not to raise prices at busy times	Whether the test went ahead, and its results

Two patterns stand out. The uses guests do not see are spreading and reported as working. And the uses guests do see have been pulled back when they got orders wrong or felt unfair.

Next: What could change next ›

Section 4 of 9

What could change next

In one sentence

Watch whether AI ordering gets good enough for busy hours, whether prices start to change by time of day, and whether AI scheduling saves money without hurting service.

Scenario 1: AI takes orders well enough for busy hours. Voice and phone ordering stop making mistakes that guests notice. Assumes: the software keeps improving, especially with noise and accents. Signal to watch: a large chain expanding voice ordering after a test instead of pulling back. Last checked: September 2026. Only 6% of operators use AI for orders, and the largest tests have been limited.

Scenario 2: Prices change by time of day. Digital menu boards and ordering apps make it easy to lower prices at quiet times. Assumes: operators frame it as discounts, not surcharges. Signal to watch: a chain testing lower prices at quiet hours without a backlash. Last checked: September 2026. Wendy's backed away from dynamic pricing after guests objected.

Scenario 3: AI schedules cut labor without hurting service. Scheduling tools become standard, and labor cost falls a point or more. Assumes: managers keep the right to override the tool. Signal to watch: published results from operators, not vendors, showing lower labor cost and steady guest scores. Last checked: September 2026. The reported savings so far come from a vendor's survey.

All three reward the same preparation: test anything guests see in a few locations, measure repeat visits as well as cost, and keep a manager in charge of the final call.

Next: Who gains and who loses ›

Section 5 of 9

Who gains and who loses

In one sentence

Operators who use AI behind the scenes gain. Those who put it in front of guests too early, or use it to raise prices, lose trust. Staff lose hours unless the time goes to service.

Gain: operators with good sales data. Their forecasts, schedules and orders get better with AI.

Gain: small operators who could not afford marketing help. AI makes posts, emails and offers cheaper to produce.

Gain: guests at quiet times, if operators use flexible pricing to offer discounts rather than surcharges.

Lose: operators who put AI in front of guests too early. Wrong orders and slow responses drive guests away.

Lose: operators who use AI to raise prices at busy times. Guests see it as unfair.

Mixed: hourly staff. Tighter schedules can mean fewer hours. Less paperwork can mean more time with guests.

Next: Where to look before you invest ›

Section 6 of 9

Where to look before you invest

In one sentence

Start with forecasting and scheduling; labor is the biggest cost you control. Add ordering and waste next, and test anything guests see in a few locations first.

Where	Prerequisite	Cost range for a group with $10 million to $200 million in sales	Most common reason it fails	Pitch you will hear
Traffic forecasting and scheduling	Clean sales data by hour and managers who can override the tool	A few hundred dollars per location per month	Shifts cut so tight that service slips	"Cut labor by three points"
Ordering and waste	Recipes and prep sheets in the system	A few hundred dollars per location per month	Nobody counts inventory, so the forecast drifts	"Cut food cost overnight"
Marketing	A clear voice and someone who checks every post	Low; often part of existing tools	Generic posts that sound like everyone else	"Never write a post again"
Phone or drive-thru ordering	A test in a few locations, with complaints and repeat visits measured	Varies widely; test before committing	Wrong orders at busy times	"Never miss a call"

The cost ranges are illustrative and will be replaced with reviewed figures. The order matters more than the amounts: start where guests do not see it, and measure repeat visits as well as cost.

One test before any contract. Run the tool in three locations against three similar ones for a quarter, and compare labor cost, food cost and guest scores.

Next: Up and down the chain ›

Section 7 of 9

Up and down the chain

In one sentence

Your suppliers, delivery apps and landlords are all changing with AI. Each change affects your costs and who controls your guests.

Food and drink suppliers. Distributors are moving small accounts to online ordering with AI suggestions. What they suggest shapes what you buy. See the guides for wholesale distribution and for beer, wine and spirits.

Delivery and booking apps. They use AI to rank restaurants and set fees. Their ranking shapes who finds you.

Guide: Social media and digital platforms ›

Landlords. Owners are using AI in leasing and pricing.

Guide: Real estate ›

Equipment makers. Kitchen equipment and building systems increasingly come with software that predicts failures and energy use.

Guide: Manufacturing ›

A good answer

A figure per tool, checked against the budget.

Push back if you hear

“The vendor's numbers.”

Next: What changed ›

Section 9 of 9

What changed

In one sentence

This guide was first published September 29, 2026. Changes that affect its conclusions are listed here, newest first.

September 29, 2026: first published.

July 2026 (reflected at launch): a vendor survey of more than 420 operators reported lower labor and food costs among AI users. Affects sections 2, 3 and 6.

February 2026 (reflected at launch): the National Restaurant Association reported that 26% of operators use AI, and only 6% use it for orders. Affects sections 1, 3 and 4.

August 2025 (reflected at launch): Taco Bell said it is reconsidering where to use voice AI. Affects sections 3 and 4, and supports the guide's conclusion: start where guests do not see it.

Next: Decision exercise ›