Drive-throughs have been a staple of American life since the mid-20th century, but how they work — and how well they know you — is about to change.
Fast-food chains are looking to deploy cameras that recognize license plates in order to identify customers, personalize digital menus and speed up sales. Starbucks began trying such a system in South Korea last year with customers who preregistered their cars. Restaurants in the United States are now looking to follow suit.
License plate recognition has been around since the mid-1970s and traditionally has been associated with law enforcement and repossession agencies. Cameras attached to police cars or lampposts read the license plates of passing vehicles and compare the results to a database of wanted cars. The system alerts officers when a suspect vehicle is spotted.
As the cost of the software, and of high-quality internet-connected cameras, has come down, the uses of license plate recognition have grown. Wary homeowner associations use private systems to spot potential criminals, and construction sites use the technology to monitor incoming traffic. Privacy advocates say that this is excessive and that the widespread use of tracking technology is dangerous.
Drive-throughs could use license plate recognition to help identify repeat customers, enabling a restaurant chain to link an individual car with a customer’s credit card and order history — meaning the customers could pay without pulling out their wallets or phones.
Customers who belong to loyalty programs or use restaurants’ apps could add their license plates to their existing profiles; cameras positioned in drive-through lanes would then take photos of cars’ plates, and the analysis software would determine whether they belonged to known customers.
How chains would ask people to opt in, and whether they would store the license plate images of those who don’t opt in, remains to be seen.
‘The advent of these capabilities’
License plate recognition start-up 5Thru said several drive-through chains in the United States and Canada were trying its technology, and it expected to sign its first major contract by the end of next year.
Chief Executive Daniel McCann said 5Thru’s technology helped restaurants process about 30 extra cars a day by reducing order time. He said the system, driven by artificial intelligence, also improves upselling by recommending items based on a customer’s past orders, the weather and how busy a store’s kitchen is.
Tracking customers using cameras is just one way stores are seeking to become more efficient in the face of online competition. Data-driven innovations include systems that alert shops when a product is out of stock and systems that try to interpret expressions on a customer’s face to gauge the person’s interest.
In March, McDonald’s bought machine-learning start-up Dynamic Yield for $300 million. Part of the idea was that Dynamic Yield, which specializes in “decision logic,” would help make food and add-on suggestions to drive-through customers who are in line. Drivers would see tailored options on digital menus, based on factors including the time of day and their selection, the chain said.
In 2017, fried chicken chain KFC partnered with Chinese search engine Baidu to develop a facial recognition tool used to predict someone’s order based on the person’s “age and mood” and recommend a meal.
Although no drive-through chains in the United States have rolled out license plate recognition at scale, McCann said, “there are a lot of conversations going on.” Jason Spielfogel, director of product management at security company Identiv, and John Chigos, founder of PlateSmart Technologies, also said the number of inquiries from retailers about license plate recognition was growing.
Meanwhile, telecom giant AT&T said it had received numerous requests from fast-food chains looking to deploy technologies such as facial recognition and license plate recognition via its 5G networks, some of which it was now working with.
“We are at the advent of these capabilities,” said Michael Colaneri, vice president of retail and restaurants at AT&T, though “nobody has quite pulled it all off.” Given increasing concerns about privacy and surveillance, he emphasized the importance of obtaining customer permission before rolling out these systems.
In addition to technical expertise, effective data-driven personalization relies on a huge amount of information about customers. Privacy campaigners have long criticized license plate recognition, calling it overly invasive and poorly regulated. In the United States, states have different rules governing the technology, including to whom these systems can be sold and how long the data may be stored.
States including Arkansas, Georgia and Maine restrict the technology’s use to law enforcement and security purposes. But business can use license plate recognition in most states without explicit driver consent: Courts have generally ruled that there is no expectation of privacy in license plates.
Although license plate recognition photos collected by police forces are protected by local laws, some vendors, such as Motorola-owned Vigilant, sell access to huge troves of such data collected by commercial customers. This information is not subject to the same usage and deletion rules that govern law enforcement.
In this context, restaurants “don’t want to talk about [license plate recognition] because it sounds too Big Brother-y,” said Aaron Allen, founder of restaurant consultancy Aaron Allen & Associates.
Which metrics are chosen to help make predictions — license plate recognition cameras can identify a vehicle’s age, make and condition — and how long to store the images remain key decisions for restaurant chains.
In 2014, a user of online forum MetaFilter asked whether McDonald’s was “running my license plate through a database, in near-real time” after being greeted with a “Welcome back!” by a drive-through employee. A debate ensued, which prompted talk of paranoia, spying, tinfoil hats and the suggestion that “scanning license plates seems like an absurd, time-consuming, expensive, and completely useless thing for a McDonald’s franchise to do.”
But in March 2018, discussing the Dynamic Yield acquisition, McDonald’s global Chief Information Officer Daniel Henry said the company could in the future use license plate recognition to personalize smart menus.
Several years earlier, in 2012, Xerox had filed a patent application for a drive-through tool to help track repeat customers, which went a step further — using “vehicle and facial information.”
Xerox has not advertised that specific product. However, the company offers license plate recognition services as well as a “passenger detection” police tool. The system uses cameras to identify a vehicle and how many people are in it, and redacts facial images “for privacy purposes.”
© The Financial Times Ltd. 2019. All rights reserved. FT and Financial Times are trademarks of the Financial Times Ltd. Not to be redistributed, copied or modified in any way.