Harikesh Nair: “The Promise of the Big Data Revolution”
A marketing professor uses analytics to help improve the bottom line at a Las Vegas gaming company.
Just a few years ago, Las Vegas gaming giant MGM Resorts International decided that its targeted marketing was, well, off target.
Like many customer-facing industries, the gaming and hospitality industry collects massive amounts of data on customers’ behavior and use of promotions. And like other casinos, MGM was interested in using advanced analytics to leverage these data to better target consumers and improve the return on investment on its marketing spending.
Professor Harikesh Nair developed analytics that help Las Vegas gaming giant MGM Resorts International better predict which customers respond to its promotions. | Associated Press photo by Jae C. Hong
So the company asked Harikesh S. Nair, associate professor of marketing at Stanford Graduate School of Business, and Boston-based consulting firm ESS Analysis to develop analytics to help it better predict which customers would respond to its promotions. The system they created involved mathematical models of consumers’ casino visiting and spending behavior, with more than 20,000 data-based parameters that showed how consumers of different profiles respond to bundles of promotions. The result: In a 2012 randomized control trial involving 2 million MGM customers, the company found that the new system brought between $1 million and $5 million of incremental profits per marketing campaign and an 8% increase in the return on investment of marketing dollars.
Previously, MGM tracked customers’ recent trips and offered complimentary food, hotel rooms, and the like to those who visited its casino frequently. But that method had shortcomings. For instance, because it used data from only recent visits, it didn’t capture a complete profile of its customers. Loyal customers who had been coming for years were likely overlooked if they hadn’t visited recently. Under the old system, MGM risked mismatching promotions to those who would play anyway, and missing some who would respond to them. It also did not provide an easy way to predict whether a customer would be profitable in the long run or suggest exactly what kind of promotion to offer what type of customer in order to maximize returns.
To sharpen the analysis, Nair and ESS created large-scale statistical models to let MGM look at more information, as well as more kinds of information. “A big part of it was improving the quality of the inputs,” says Nair. The new model, for instance, reflects a longer history of visits, a finer way of segmentation, and a fuller picture of the set of promotions available to a customer, giving MGM a more complete picture of each consumer and his or her decision calculus. This, in turn, enables a more sophisticated way of understanding how promotions affect behavior.
The new model also provides analyses even in instances when MGM has little data about a particular customer. It examines information about others who are demographically similar to try to predict that customer’s future behavior and profit potential. For example, if a customer visited MGM only once and spent little on the trip, the model looks at the long-run spending of others similar to that customer. If the others spent little on the first visit but dropped a bundle on subsequent trips, the system will target the customer in question even though he or she spent little the first time around. “This person could be profitable from a lifetime perspective” and might be a candidate to receive a promotion, says Nair.
Through analyzing a combination of past behavior and other traits, MGM can now identify, for example, customers whose likelihood of a future visit depends on their winnings during a previous visit. This can be useful for designing loyalty programs and for understanding how to maintain a customer relationship over the long run, Nair’s paper on MGM notes.
MGM’s experience suggests that other companies, ranging from movie theaters to clothing retailers, could also increase their profits and ROI through better use of data collected through loyalty programs, surveys, and other means. “In those environments, massive amounts of data are available,” says Nair. “Similar kinds of models as developed in this study may be developed for those settings as well.”
Nair notes that MGM’s experience shows the value of using marketing analytics and statistical methods on massive volumes of customer data. “Even a small improvement in analytics can vastly improve your ability to make better decisions,” he says. “Good analytics, combined with great data, complements smart management. This is the real promise of the ‘Big Data’ revolution.”
Harikesh Nair discusses his work with MGM in the case study “Big Data and Marketing Analytics in Gaming: Combining Empirical Models and Field Experimentation.” The paper, which hasn’t been published, was co-written with Sanjog Misra of UCLA’s Anderson School of Management, Ranjan Mishra and Anand Acharya of ESS Analysis, and William J. Hornbuckle IV, president and chief marketing officer of MGM. For this study, MGM retained ESS Analysis, which in turn paid Nair as a research adviser.
For media inquiries, visit the Newsroom.