Why the Airline Industry Needs Another Data Revolution

Over the years, airline travel has been a prime testbed for advanced computing and data tools. In the late 1950s and 1960s, American Airlines and I.B.M. teamed up to develop the Sabre computerized reservations system, perhaps the most impressive private-sector computer system of its day.

More recently, airline data has served as the raw material for predictive data-mining applications like Farecast, which tells consumers whether the price of a plane ticket, for a specific trip on a specific day, is likely to rise or fall. (Farecast, founded in 2003 by Oren Etzioni, a University of Washington computer scientist, was sold to Microsoft in 2008. It is now part of Bing Travel.)

But the airlines themselves have become laggards in data-handling innovation, according to Thomas H. Davenport, a visiting professor at the Harvard Business School. “They were early adopters, and they have not done much for many years,” Mr. Davenport said. Their loyalty programs and per-s eat revenue management systems, he said, both date to the 1970s.

Mr. Davenport is a longtime expert on information management and data analysis. From his main perch at Babson College, he has been researching and writing about the field for a few decades, tracking the evolution of the technology and the terms used to describe it - from business intelligence to analytics to Big Data.

Mr. Davenport, co-author of a new book, “Keeping Up With the Quants” (Harvard Business Review Press), with Jinho Kim, does his research on the quantitative world the qualitative way. He interviews people and does case studies.

His comments on the airlines result not from his work for the book, but from research for a study published on Wednesday, “At the Big Data Crossroads: Turning Towards a Smarter Travel Experience.” His research was sponsored by Amadeus IT Group, the big European computer reservation system and technology services company. But Mr. Davenport said it did not guide his research. (“We're just trying to facilitate the debate,” said Hervé Couturier, an executive vice president of Amadeus.)

For the 28-page report, Mr. Davenport interviewed executives at 21 companies involved in one way or another in travel, but representing a cross-section of airlines, hotel chains and technology companies. The companies included Air France-KLM, Applied Predictive Technologies, Facebook, Hipmunk, Intercontinental Hotels and Marriott.

The report includes short case studies on companies that are doing innovative data projects. Mr. Davenport's exemplary airline is British Airways for its new personalized service and offers program, Know Me. It collects and tracks an usual amount of data on individual passengers, their preferences and travel history.

If a person's bag went astray on a flight, that individual might be offered a free upgra de for his or her next flight. The system has the ability to identify customers and instantly suggest tailored offers at check-in counters or lounges. On planes, service personnel with iPads can make authorized offers for custom services. “If it's really personalized and appropriate for the context, it can be seen as a service instead of a marketing program,” Mr. Davenport said.

The software technology behind the Know You program is supplied by Opera Solutions, a New York-based Big Data analytics company. It assembles data on the online behavior and buying habits of 20 million British Airways customers, creating hundreds of predictive signals that suggest a person's “behavioral DNA,” Arnab Gupta, chief executive of Opera Solutions, said in an interview.

Such signals, he said, might include a person's tendency to book an airline ticket a month or more in advance or buy a ticket a few days ahead. Other signals might be a person's history of booking two-star versus five-star hotels. Online behavior might include visits to BA.com and whether a person booked with a few clicks or frequently abandoned digital shopping carts.

A key conclusion from Opera Solutions' work with companies in many industries, including travel, Mr. Gupta said, is that “90 percent of the predictive value is in the behavioral data.” That is, by monitoring what people do online and in the physical world rather than demographic profiles that seek to predict what people will do based on gender, race, age and income.

“It's liberating,” Mr. Gupta said. “We cluster more by behavior than by demographics.”