Features | From Pivot Magazine

The newest, cheapest way to book a plane ticket: artificial intelligence  

Hopper wants to be your new favourite travel agent. Can it compete with established flight trackers or will the competition carry it away?  

A Facebook IconFacebook A Twitter IconTwitter A Linkedin IconLinkedin An Email IconEmail

photo illustration of multiple and different jet planes in the blue sky Launched in Montreal in 2015, Hopper is one of the first flight-booking applications to use artificial intelligence to predict the cost of flights (Planes by iStock and Alamy)

Over the past decade, machine learning has infiltrated nearly every imaginable sector. Art­ificial intelligence helps doctors diagnose patients, researchers develop new drugs and bank customers manage their cash flow. Now, the AI revolution has reached a new frontier: the travel business. 

Launched in Montreal in 2015, Hopper is one of the first flight-booking applications to use artificial intelligence to predict the cost of flights. The process is relatively straightforward. Once a user downloads the Hopper app and inputs a destination, they are presented with a year’s worth of colour-coded calendar pages. Green represents the cheapest flights, two darkening shades of orange are less so, and red is caviar territory. The app’s algorithm churns through Hopper’s database, which contains several trillion price points harvested over the last several years, then spits out a price—and, often, advice not to pay it.

“You should wait for a better price,” reads a Hopper app notification after delivering a return price of $459 from Montreal to Orlando in September. “You could save as much as $133 per ticket by watching this trip.” Hopper then sends out push notifications advising the user to either hold off or take the plunge. It will also suggest cheaper alternatives and send notifications if there is a sudden price drop.

Hopper’s algorithmic nudges are increasingly popular. The brainchild of Frederic Lalonde, a proud Quebec City-born college dropout and hacker, the app has more than 40 million downloads, with another million downloads every month. It had $1 billion in sales in 2018 and has secured $235 million in venture capital funding since 2015. With 289 employees in Montreal, Boston, New York and Sofia, Bulgaria, Lalonde hopes to double the number of tickets booked via Hopper by the end of 2019. He has an even more ambitious plan for his company: to become a “super app” alongside expected trip-booking offerings from giants Amazon and Google, which would combine flights, hotels and other travel reservations in one place.

portrait of 'Hopper' founder Frederic LalondeHopper founder Frederic Lalonde (Courtesy of Hopper)

Hopper recently began applying its “hurry up and wait” model to hotels, allowing users to gauge the best time to book a room. Even the smallest of toeholds in the industry would be lucrative: travel and tourism contributed $11.7 trillion to the world’s economy in 2018, according to the World Travel and Tourism Council.

Yet competing with well-established, deep-pocketed behemoths remains a massive challenge for the company, which has neither the size nor the advertising budget typical of the leading industry players. Meanwhile, its hometown of Montreal is stymied by a shortage of programmers and other skilled employees, adding pressure to an already fraught question dogging many small, successful startups in Canada: pivot to greatness, or sell out and start over?

For air travellers, 1996 was a game-changing year. Until then, the freedom-inducing prospect of flying several hundred kilometres an hour at 30,000 feet was hampered by the all-powerful travel agent. Booking a flight meant going through these gatekeepers, a process that often involved a fax machine, the postal service and associated aggravation. Expedia, launched in 1996 by the then-almighty Microsoft, democratized this decidedly bureaucratic process by allowing anyone with a desktop and a dial-up modem to book a flight with their fingers. Between 2006 and 2016, nearly 860 billion kilometres of air travel were booked through Expedia.

For the most part, the revolution stopped there. Today, there are countless other websites offering flights, hotel rooms and rental cars, as well as aggregators that sift through those sites to spit out the lowest fares. Many of them are older and more established than Hopper: the Connecticut-based Kayak, for one, went public in 2012, and the Scottish flight-search service Skyscanner was acquired by Chinese travel giant Ctrip for $1.75 billion (U.S.) in 2016. But Lalonde argues the process on those sites—plug in destination and date, cross fingers, input credit card information—is the same as it was the year the Motorola StarTAC was born. “You can still hear Pearl Jam playing in the background when you look at these sites,” says Lalonde.

“It was the typical startup story: trying to convince a giant company to do something that made no sense at the time.”

Lalonde came by big data, the industry’s potential disrupter, unconventionally. The 46-year-old, who favours V-neck shirts and sculpted stubble, grew up a hacker. In the early days, when video games cost $20 and came on floppy disks, Lalonde would crack their security protections, download their data and duplicate them. He’d sell bootlegs of games like Boulder Dash and battle-chess staple Archon in the schoolyard for five dollars. “For a 14-year-old, I was making some really good money,” he says.

Lalonde finished high school but bridled at CEGEP, Quebec’s system of post-secondary, pre-university schools. He dropped out and found himself in a beautiful but decrepit loft space in Old Montreal, where he and partner Benoit Jolin contemplated the sheer inefficiency of the world’s various hotel reservation platforms. Somehow, booking a hotel room in the year that brought us Titanic was more difficult than getting to it by airplane, with reservations spread out over multiple systems that often didn’t communicate with one another. In 1997, Lalonde and Jolin founded Newtrade, which essentially linked and computerized the various silos. They secured a then-hair-raising $7 million in funding and employed 70 people to do what Lalonde called “the really crappy dirty work” of populating their computer model with data from the hotel industry—a back-end process that few nascent internet companies were doing at the time. 

In 2002, following a chance encounter with Expedia executive (and Montreal native) Erik Blachford, the travel site acquired Newtrade. The sale brought Lalonde into the Expedia fold as a vice-president, where he worked under the founding CEO, Richard Barton. For a dropout like Lalonde, the job was like going to school. “Everything stems from Barton’s vision,” Lalonde says today. “His thesis is to take a consumer category and find exclusive, opaque data that nobody knows or that only a few experts have access to.” Build a company around that, he says, and they will come. “First, find some high-anxiety consumer problem, and then use data to fix it.”

By 2006, Expedia’s competitors had started offering variations on—but not significant departures from—the Expedia experience. With Barton’s mantra in mind, Lalonde left the company, turning his thoughts again to potential sources of untapped data. He found the motherlode in the Global Distribution Systems, the clearing houses for the majority of the world’s flight reservation information. Specifically, Lalonde wanted “shadow traffic” data revealing what people were searching, how many people were doing so, when they were travelling and what prices they were quoted. In 2007, when he and former Newtrade software architect Joost Ouwerkerk founded Hopper, precious few in the industry could figure out why on earth he wanted any of it.

Lalonde remembers approaching airlines, requesting data and offering nothing in return. “You can imagine it took some time to cut these deals,” Lalonde says. “It was the typical startup story of trying to convince a giant company to do something that made no sense at the time. But today, we send tens of millions of dollars of business to these companies. The guys that cut those deals are rock stars now.”

“Fundamentally, customers don’t wait for the last price drop. Nobody waits six extra days for five dollars.”

Over the course of several years, Lalonde and Ouwerkerk began hoovering and stockpiling the data, often upwards of 30 billion prices a day. Patrick Surry, Hopper’s chief data scientist, believed he could make accurate predictions as to the price of airline tickets using machine learning and three years’ worth of data. The company sold 1,000 tickets in the app’s first week of operation in 2015, saving customers up to 20 per cent per ticket. Today, with tens of trillions of data points feeding its algorithm, Hopper claims to be able to predict prices with 95 per cent accuracy. 

“Fundamentally, customers don’t wait for the last price drop. Nobody waits six extra days for five dollars,” he says. “As soon as they feel like they’re doing okay and they haven’t overpaid, they pull the trigger and they move on.” He says 90 per cent of flights booked on Hopper come from the billions of push notifications the company sends out advising people to buy a ticket. “It works because you do no work and you save money. It’s the perfect consumer-value proposition. It’s effortless.”

Yet some experts doubt sites like Hopper can reach past their base of budget-minded customers and appeal to business and luxury travellers, considered the industry’s gold mine. “It’s a very dynamic application,” says Rodrigue Gilbert, managing director at PwC Canada, where he is also lead relationship partner for transportation and logistics. “But it won’t dominate the market. It only appeals to a small fraction of customers who are more price-sensitive.”

Hopper will also invariably face competition from those established travel sites, all of which have their own data to mine—and whose existing footprints are exponentially larger than Hopper’s. “The big guys will react for sure,” says Gilbert. “They have the data, but they don’t have the engines to put it to use yet.” He says a number of airlines are currently working with AI labs to analyze customer preferences and choices.

Hopper has two things the bigger players don’t have: a head start and a fiercely loyal customer base

Finally, there is the issue of funding and skilled labour, both of which are in shorter supply in Canada than, say, Boston, where Hopper already has a large presence. Will Hopper fall prey to the curse of the Canadian startup and sell out before it matures?

Montreal, like much of Quebec, is a victim of its own success. A rollicking economy, coupled with a decades-old reputation as a technology hub, has meant a boom in startups—but a chronic dearth in people with the skills to work at them. A 2018 government report forecast a continued labour shortage in the sector across most of the province. Such a shortage limits growth potential for fast-developing companies like Hopper. No coincidence, then, that Hopper opened sizable satellite offices in Boston and New York, travel hubs with comparatively large pools of skilled IT talent. 

Lalonde says Montreal’s universities should emulate the University of Waterloo model: create a collaborative, city-wide tech network that encourages students to work at local companies after they graduate. “We don’t have co-op programs in Montreal, at scale, at any of the tier one universities,” he says. “If we replicated the Waterloo model nationwide, we would actually start creating one of the richest factories for talent, because people want to live in Canada.”

Unsurprisingly, Lalonde maintains Hopper isn’t budging from Canada. As evidence, he points to funding from Investissement Québec and Quebec’s Caisse de dépôt et placement, the province’s pension manager, both of which tend to favour long-term investments that are likely to benefit the province. And according to PwC Canada assurance partner Andrew Popliger, Hopper has two things the bigger players don’t have: a head start and a fiercely loyal customer base. 

“We’ve had offers to sell that we’ve turned down,” Lalonde says. “At some point, as a CEO, you have to decide what company you’re building. Are you going to flip it? Are you going to go public quickly? Are you going to hold for the long run and go really, really big? And we chose the third option. We’re going to go really, really big.”