Only 3.2% of restaurant operators run AI companywide, NRA survey finds

Restaurants have spent about 25 years adding technology one problem at a time, and 82.2% of operators say it has helped their business. AI is the newest layer. The National Restaurant Association's new paper, Research Insight: Digital Innovation, The Future of Restaurant Technology, published November 5 with DoorDash as presenting sponsor and PayPal as supporting sponsor, shows how thin that layer still is.

Nearly four in ten operators use AI, but only 3.2% run it companywide, which puts most of the industry somewhere between a test kitchen and a rollout. The paper's argument, which the Association repeated in a November 10 article, is that the next gains will come from connecting the ordering, payments, labor, inventory and marketing systems restaurants already bought and turning their combined data into faster decisions on staffing, prep and purchasing.

The Association aims the paper at multiunit operators and growing restaurant companies. For the operations or IT lead at one of those chains, that changes what to ask before the next technology budget. Asking which tool to add next matters less than finding which existing systems still can't share data.

Wide use, thin rollout

Operators who use AI mostly point it at personalized marketing and customer feedback analysis. Put the companywide rollouts and the pilots together and you get roughly 23%, which is our arithmetic on the survey figures. That falls well short of the 38.6% who say they use AI tools at all. The difference suggests that a good share of restaurant AI today is software aimed at marketing and reviews, not a structured project tied to forecasting or labor.

The intent is clearly there. More than three-quarters of respondents, 77.9%, expect technology, automation, data and AI to be at least moderately critical to their businesses over the next five years. Interest outruns execution, and that gap is the useful part of this survey.

The method matters for how much weight these numbers can carry. According to the Association's release on PR Newswire, it surveyed 212 operators between June 18 and July 18, 2026, and interviewed eight leaders across quick-service and full-service restaurants. Consumer figures come from its 2026 State of the Restaurant Industry report. At 212 respondents, the percentages are a directional benchmark for internal planning. They are not precise market sizing.

Barriers that live in the org chart

Asked what stops them from adopting or expanding technology, operators put cost first. The next three answers have more to do with how the business runs than with what the software can do.

The leaders interviewed for the paper said organizational factors were bigger obstacles than technical limits. They described AI plans that run ahead of the quality of the data behind them, and said training and change management matter as much as the tools. Chad Moutray, the Association's chief economist and vice president of research, put it bluntly: technology can't make up for inconsistent processes, unreliable data or unclear goals, but it can amplify a sound operation.

Operators seem to agree. Training and education, at 56.5%, topped the list of support they said would help them adopt these tools. Nearly half (47.6%) expect technology and AI to raise labor productivity, and 37.2% expect employee roles to require more technology skills. For a chain planning an AI rollout, the training budget belongs right next to the license fee.

The margins explain why. Restaurant profit margins typically run 3% to 5%, and 96% of operators call labor costs a significant challenge, with 95% saying the same about food costs. Those figures come from the Association's 2026 State of the Restaurant Industry data, as cited in a September blog post by NCR Voyix. With margins that thin, a slightly better prep and staffing forecast can be worth more than another dashboard.

Connected data in a working kitchen

The restaurant leaders in the paper gave two concrete examples. Some now produce demand forecasts for individual products at individual stores. Others use AI to answer questions in leadership meetings that used to take days of analysis. The paper sorts the value into three areas. Integrated sales, labor, inventory and customer data improve forecasts and speed up staffing, prep and purchasing decisions. Loyalty, ordering and reservation data support better-targeted offers and recognition of special occasions. Automation cuts paperwork so managers have more time to coach staff and serve guests.

Hotel food and beverage teams show what this looks like in practice. In October, Elyse Glickman reported in Hotel Management on Dallas's Hôtel Swexan, which runs five restaurants, a banquet floor and thousands of SKUs. According to David Cantu, CEO of back-office platform Craftable, the hotel's monthly inventory close fell from six days to one after it adopted Craftable. His explanation is the integration problem in miniature. Purchasing, invoicing, inventory, recipe costing and labor planning often sit in tools that were never built to work together. So when food costs move, recipe pricing doesn't change until a manager makes the connection by hand.

Concord Hospitality took a different route and built its own tools, including one called Yes Chef plus custom Copilot and Claude integrations. Corporate F&B director Zachary Wendel described the aim as a single source of truth across outlets ranging from rooftop bars to full-service restaurants. Concord still spot-checks pricing, which is worth noting: connected systems cut manual work, but a person still has to own accuracy. In the same piece, Solink CEO Mike Matta warned that disconnected tools tend to become one more dashboard to check, when teams need answers they can act on.

The hospitality test

The paper measures all this efficiency against the guest experience. Sixty-five percent of operators say technology improves the hospitality they provide. On the consumer side, technology options influence restaurant choice for 77% of delivery customers, 66% of takeout customers and 65% of drive-thru customers. About 7 in 10 consumers say they would order and pay through an app, website or kiosk if one were available.

That does not mean guests want automation everywhere. Customer acceptance ranks third among operators' AI concerns, at 41.9%. A guest happy to tap a kiosk may still be uneasy about AI shaping the rest of the visit. The connected-tech case holds only where the staff time it frees shows up somewhere the guest can see it.

Pepper Lunch is one example. Former CEO Troy Hooper described the setup at the MURTEC Executive Summit 2026: kiosks take orders, and hospitality hosts welcome guests, explain the menu and step in when someone needs help. The technology handled the order. People were given the job of handling the guest.

Checks before signing the next contract

Chris Munz, a fractional CTO and restaurant technology consultant, wrote up the MURTEC summit in Coronado, California, for Hospitality Technology. Several of the tests he described line up with the barriers in the Association's data. On integration, he suggests following one guest preference change from the app through to the email program. Find out where the preference lives, who can see it and who fixes it when it doesn't carry over. On build versus buy, give current vendors a real scenario and have them walk through it with the team. If the chain decides to build, name the person who will own the integration when it breaks six months after launch.

For a first AI project, he suggests asking GMs what they still do by hand at the end of a shift. Pick one of those tasks and time it before and during the test, including the time spent checking or correcting the output. Deric Rosenbaum, speaking on a MURTEC panel about connected data, described putting prep guidance on the screens kitchen teams already use, then tracking what happens to waste and stockouts.

The number to watch is the 20% of operators running small pilots. If a meaningful share of them move into the companywide column in the Association's next reading, the integration and training work is paying off. If the pilot share just keeps growing, the industry is mostly buying more tests.