Too many big data initiatives are science projects that take months of effort, risk failure and require highly trained data scientists with scarce skills. According to a CSC survey, 55 percent of big data projects aren’t completed and many others fall short of their objectives.Read more...
How Big Data Helps Airline Profitability
The advent of big data technologies is impacting companies in all industries. But in the $743 billion global airline industry, big data analytics is increasingly the difference between successful airlines and those who struggle to prosper in an increasingly competitive field.
The airline industry expects to make a profit of nearly $20 billion in 2014, up about 50 percent from 2013 profits, according to figures released last month by the International Air Transport Association (IATA). That is an astounding turnaround from 10 years ago, when nearly half of the biggest U.S. airlines were in bankruptcy, and the industry was losing about $10 billion per year.
To be sure, there are many reasons for the improvements to the airline industry. Big mergers have reduced the number of airlines, and the remaining airlines have taken planes out of service, thereby boosting demand for the lower number of remaining seats. Fuel prices have eased somewhat since peaking in 2008. And the Great Recession has thwarted wage growth for workers, which is another big cost for airlines.
But there’s something else going on in the airline industry, and some call it big data. The most technologically progressive airlines have figured out how to use emerging data analytics technologies to boost sales and improve razor-thin profit margins. With profits hovering around $6 per passenger and projected to drop this year, airlines are under the gun to do anything to improve the situation.
Tim Simmons, Teradata’s vice president of global industry marketing for the retail, travel, and transportation sectors, identified three major ways that airlines are using big data analytics to their advantage. The first two–capturing sensor data to optimize maintenance and forecasting the weather to optimize fuel loads–are basically iterative enhancements upon activities that airlines have been doing since the dawn of the commercial flying age.
The third big data category–identifying and capturing the demand signal–offers much greater opportunity for airlines to differentiate themselves. This is where airlines are looking to use emerging technologies, such as Hadoop and sophisticated data mining algorithms, to capture unstructured data.
“That’s probably the single biggest opportunity, which is merging traditional warehouse with the new unstructured data around sentiment and blogging,” Simmons tells Datanami. “What the airlines are trying to do is to build a 360-degree view of the customer to understand and append on top of the easy-to-understand structured data. They’re now trying to add on the unstructured data, which is more to do with sentiment.”
The advent of Hadoop is allowing airlines to embark upon massive data collection programs. The airlines are collecting customer sentiment data and ecommerce behavior from popular social media websites like Facebook and Twitter and travel websites like Kayak and Travelocity. More often than not, this data gets parked in Hadoop, where they may use Teradata’s Aster platform to run MapReduce analytical workloads to determine the level of correlation among different variables, such as ticket prices, baggage fees, route preferences, equipment failures and delays, and in-flight food purchases and entertainment.
“Companies are installing discovery platforms so they can experiment…and learn more things about their customers and networks,” Simmons says. “They say, ‘I have a hypothesis that there might be a correlation between equipment failure and customer loyalty. Can you show me?’ Once you show there is a correlation, then you throw it over the fence and it becomes a report.”
The airlines industry is in the midst of a transition right now between the early adoption phase and mainstream adoption of big data technologies, Simmons says. “Two years ago, it was possible but not practical,” he says. “Now it’s practical. And that’s, of course, why it’s carried out at scale.”
Airline’s data programs often grow out of existing customer loyalty programs. But increasingly, airlines are widening and expanding their information gathering programs to cover all airlines passengers, whether or not they have ever flown with a given airline. Only about half of passengers on any given plane belong to the loyalty program of the airline they’re flying, Simmons says.
Qantas is leading the charge in this regard, and is now actively pursuing customers on social media sites. “Qantas is leading the way with a very interesting concept in its loyalty program because they’re now looking outside at new retail tie-ins that could be relevant to the majority of their customers,” he says. “They’re a very forward thinking airline.”
Airlines are also starting to use Hadoop to store vast amounts of its customers’ Web traffic data to get a better idea of how its customers made their selection. By grabbing the cookies that each Web user brings with them when they visit an airline’s e-commerce website, an airline can find out what other sites a given consumer visited and what other searches they performed before purchasing a ticket.
If and when the consumer buys a ticket and identifies himself through his payment, an airline can “go back in time” in effect and see what other sites he visited. This helps the airline to build a better model of customer behavior, and identify which advertising, email, or pay-per-click marketing programs are effective and which aren’t.
“That’s actually a very common behavior now for businesses whose revenues come from digital channels,” Simmons says. “It’s not a new technique. The database technology itself is making it more possible to track more and more different kinds of cookies. Historically we could just track cookies from your main point of entry, which was your PC. But now we can correlate cookies from mobile device as well, which is very helpful.”
Attribution and path-to-purchase analysis used to be difficult and expensive analytical tasks best left to senior data scientists. Simmons argues that, by making these functions available as pre-built algorithms in its Aster platform, that Teradata is lowering the bar of entry. “So an ordinary analyst, who doesn’t have to be data scientist, can kick off a very sophisticated query and get a result set back that has huge value to an organization.”