Authors’ Note: In our last article ("Let’s Talk Turkey: Applying Retail Market Basket Analysis to Gaming," in the December 2008 issue of Casino Enterprise Management), we discussed the rudiments of the retail industry’s market basket analysis theory and provided a simple example of how it can be applied to the gaming industry. We also pointed out that the nature of casino products differs from grocery store products in that the customer chooses the price (e.g., multi-denomination slot machines) and receives the same product. In this article, we will provide more-sophisticated and practical illustrations, identifying a particular customer interaction with a slot machine as a market basket product. This work will be extended in a third article that will show how customer demography can be combined with the market basket to enable marketing activities that combine both behavioral and demographic insights.
Previously, we looked at a small data set from a small casino that had two categories of slot machines, a table game, and a hotel. Although some interesting conclusions could be drawn from the simple example, market basket analysis (MBA) really shines when dealing with data sets that have a large number of cases and variables.
In the world of flexible games with multiline/multigame options—and the future world where games themselves will be changed dynamically—the definition of a particular slot machine may incorporate numerous attributes. One example is a measurement of which bet is most commonly placed on a multi-denomination slot. Thus, the reverse view of the MBA is that it not only correlates "hard" attributes of the gaming device but also the measurements of how they are played. This may result in the same physical game being classified in different ways, depending on how the customers interact with the game and the remainder of the basket.
In this scenario, imagine you are Terry Benedict (of Ocean’s Eleven fame). You run a large Las Vegas casino, and you need to figure out how to recoup the money recently stolen from your vault. You’d like to do this through more casino revenue in the high-roller slot area, where you have equal numbers of slots in three denominations: $25, $200 and $250. Despite traditional thinking, you believe that the slot players in the high-roller area are not as sophisticated as those on the main floor; you also believe that you can make more profit by leaving the hold percentage the same whether or not a player plays the max bet or not. Currently, the casino’s advantage is 10 percent when the maximum bet is not played and 5 percent when it is. The question is: Are those who play maximum bet sophisticated players who only play maximum bet, or does the "market basket" include both maximum and non-maximum bet play? This example shows how information regarding purchasing or playing behavior, or product variations, may be included in MBA by treating these attributes as another product. Table 1 contains a sample of the hypothetical data we created on player visits to the high-roller slot area.
We might conclude from this hypothetical analysis that: (1) $25 dollar players are possibly a unique segment and that it may be profitable to provide them with unique offers based on this behavior; (2) it may be possible to increase the hold percentage on $25 dollar machines when max bet is played and not affect overall play; and (3) doing so will probably not affect other non-max bet players’ play.
Given the conclusions drawn from the MBA performed above, we now want to explore whether or not we can further split the $25 market segment into profitable and less-profitable players—and then send then attractive offers. We believe that $25 players are generally not affected by the difference in the machine hold between making maximum bets and non-maximum bets. Let’s see if, in fact, some of these players are more profitable than others by looking at actual win instead of the traditional theoretical win analysis.
The market basket can be extended to include other areas of business by applying the same principal of combining the spending patterns in gaming with the products purchased in non-gaming areas. In this way, we approach a 360-degree view of the customer.
For this analysis, we will employ what we call "visual data mining." We present data in such a way that the mind can find useful patterns and information. In this example (see Figure 1 at right), we show an interactive MBA system where a basket made up of afternoon sales ($70 average) from many casino areas is examined. The green lines show the connection to the market basket components, namely the spa, the hotel, a slot category, and horse racing.
The key to this visual basket mining is that it enables interactive exploration of the complex combination of actual product choices made by customers during a short time period. By looking at the selected ad and highlighting "70 basket," we understand immediately that those customers play a balanced amount of the slot and table products analyzed but spend much less on horse racing and non-gaming products. Therefore, marketing programs targeting this group should be focused on gaming incentives (e.g., bonuses for gambling during the afternoon or tournaments).
In this exercise, MBA led us to not only discover potentially profitable operational policies (increasing the hold), but it also assisted us in identifying new market segments and their characteristics. This understanding will help us to target appropriate messages to the right audience at the right times. By applying these techniques to enough situations, Mr. Benedict will be well on his way to refilling his vault.
Bart Lewin has more than 25 years of experience in the Engineering and Information Technology field, holding serveral technical and executive technical management positions. He is currently a technical and management consultant.
Dr. Ashok K. Singh has taught statistics, mathematics and operations research courses at New Mexico Tech, Socorro, N.M., and statistics and mathematics courses at University of Nevada, Las Vegas. He has over 75 publications in theoretical and applied statistics.
Andrew Cardno has more than 16 years of experience in business analytics, ranging from modeling healthcare drive times to casino gaming floor analytics. He often presents on the future of analytics across the world and has spent the last seven years living in the United States and working with corporations around the world.
Paul Cardno has the rare distinction of being an accomplished programmer, scientist and artist. He holds a Bachelor of Science degree (with honors) in Physics, a Master’s degree in Medical Physics, and has also studied toward a degree in Fine Arts.
Comments
Post new comment