Reference no: EM131040433

Predicting Customer Behavior

Seminole Gaming (www.theseminolecasinos.com) operates seven casinos in Florida on behalf of the Seminole Tribe. Two of the facilities are Hard Rock-branded hotel casinos. The company has more than 11,000 slot machines, 300 table games, dozens of restaurants, and nearly 10,000 employees. Despite the breadth of its operations, however, Seminole Gaming considered its customer analysis to be inadequate. The company's customer segmentation strategy relied entirely on prior customer behavior, which it analyzed using traditional RFS (recency, frequency, spend) metrics.

This system did not enable the company to predict customer behavior, or as they put it, "to be able to see into the future." To achieve this objective, the company decided to implement SAS (www.sas.com) Enterprise Miner, a business intelligence package. Seminole generates a large volume of direct mail. Whereas many other industries employ newer electronic channels such as e-mail and social media as marketing tools, Seminole maintains that in the casino business, direct mail is a much more effective channel. To make this strategy more effi cient, Seminole needed to eliminate the cost of sending mail to prospective customers who are unlikely to respond. Consequently, Seminole utilized Enterprise Miner to build an analytic model for a direct mail campaign for concerts at Hard Rock Live, located on the grounds of the Seminole Hard Rock Hotel & Casino in Hollywood, Florida. The model helped the company identify the 35 percent of its customers who were most likely to respond to one of their direct-mail offers. Seminole sent each of these customers a specifi c mailing for each concert. For the remaining 65 percent, the company consolidated mailers-that is, it advertised multiple concerts, instead of a single concert, in each mailer.

While the single-mailer-per-concert approach proved most effective for those customers who were likely to respond, advertising multiple concerts in a single mailer to the other customers not only reduced the company's mail costs, but it increased the customer response rate. Overall this new response model generated more than $1 million in profi t annually. After the casino identifi ed people to whom they should not be mailing, its next strategy was to identify prospective customers to whom they should be mailing, but were not. This strategy is important for companies that rely heavily on direct mail. Seminole also utilizes Enterprise Miner to gather a wealth of data about its customers' purchases.

A signifi cant portion of its player base uses its casino rewards card. For each of these players, the casino knows their play choices as well as the outcomes for every machine or table on every day. Traditional casino direct-mail programs rely heavily on metrics such as average daily actual (a measure of how much money a player loses on a given day), average daily theoretical (a measure of how much a player would have lost if he or she had been either more or less lucky than expected, per day), average daily worth (a calculation that combines the previous two metrics), and points earned (a measure of how much total play a player has given the casino). It is unprofi table to mail every customer for every marketing campaign, so casinos use these metrics to decide to whom they will mail , and how often. Let's look at an example. Suppose Seminole launches a campaign that offers $5 in free play to any customer whose average daily worth is greater than $50. In effect, the casino has determined that offering $5 to such customers generates enough additional casino visits to cover the expense of the offer to customers who were going to visit anyway.

The question now becomes: How can the casino fi nd the "hidden gems" among the customers whose average daily worth is less than $50? The answer is to go back in time and look at all customers whose value prior to time period X was less than $50 but who came back after that time period and increased their daily worth to more than $50. This is a basic change-of-behavior model that Enterprise Miner handles easily. In fact, models such as these are driving strong, profi table results at Seminole. Seminole Gaming also engages in projects that might not generate an immediate direct return. One such project is to perform market basket analyses of the slot machines that their customers play. Market basket analysis is a technique that identifi es co-occurrence relationships among activities performed by specifi c individuals. As a simple example, suppose a casino had only three slot machine games-A, B, and C-and two customers-Bob and Rita. Bob and Rita each spend $100 per visit, but while Rita plays only game C, Bob splits his money between games A and B. In this example, the casino would say that games A and B are associated with each other, whereas game C is not associated with any other games. If you add millions of customers and thousands of slot machines to the analysis, the model becomes extremely complex. Fortunately, Enterprise Miner is able to process these huge models.

When Seminole initiated this project, they did not know what they would fi nd. The analytics ultimately revealed surprising relationships among games on the slot fl oor. The company would not publicize those relationships, but it conceded that it revised its entire slotmachine decision-making process as a result of these analyses. In its next project, Seminole is analyzing slot-machine data to uncover unknown groupings of customers and slot machines. For example, where should slot machines be physically located in the casino to attract the most customers. As of mid-2013, the results were not defi nitive. Nevertheless, Seminole is confi dent that it has established an entirely new way to segment customers, as well as an entirely new model for locating its slot machines inside its casinos, as a result of the hidden groupings. And the bottom line? In terms of benefi ts, the models that Seminole has utilized for direct-mail marketing campaigns are providing an annual benefi t of more than $5 million. Source: Compiled from R. Thomas, "Analytics Tool Predicts Customer Behavior," Baseline Magazine, August 3, 2012.

Questions

1. What other analyses should Seminole perform? Provide examples to support your answer.

2. Are there disadvantages to Seminole's use of analytics to predict customer behavior? (Hint: Is it possible to know too much about customers from the casino's viewpoint? What about from the customer's viewpoint?)