Authors’ Note: This article, the third part of a series on the mathematics of gaming analysis, takes a look at the effects of social networks within a gaming floor. These invisible forces are a combination of actual social networks and subconscious behavioral patterns.
This is the third in a series of articles on gaming floor analysis; in these articles we use mathematical models to provide deeper insight into the factors that drive gaming results. This article establishes the analytical building blocks that can be used to decode how social networks influence game play. This article builds on the first article in the series that discussed how fuzzy spatial association rules and gravity modeling can be combined for analyzing casino floor data, and the second article that explained the difference between expected value of a game and player’s gaming experience. There will be further articles on the use of experimental designs and advanced statistical methods in casino management.
Social Networks: The Mass Personal Communication
Social network analysis is a hot topic in the business world today. We only have to look at the influence of Twitter, Facebook and LinkedIn to see how people interact today using social networks. In fact, Peter Yesawich Jr., a leading interactive media consultant, recently wrote an ACEME blog on the topic that’s worth reading (visit www.aceme.org/blog/twitter-most-talked-about-misunderstood-social-media-…).
One uses social network analysis [SNA] to map and measure the relationships and flow of information between people, groups, organizations, computers, URLs and other information/knowledge entities connected through a network. The nodes in the network are the people or groups of people, while the links are used to show relationships or flow between the nodes. SNA is used to provide both a visual and a mathematical analysis of human relationships. (Note: SNA is used in the business world where it is referred to as Organizational Network Analysis or ONA).1
These interaction networks present a huge wealth of behavioral information. They also represent a sea of change when it comes to how people and organizations communicate these days. In the past, communication was either mass media or personal, but now we have a new method, which can be called “mass personal.” In fact “research in a number of academic fields has shown that social networks operate on many levels, from families up to the level of nations, and play a critical role in determining the way problems are solved, organizations are run, and the degree to which individuals succeed in achieving their goals.”2
In this world of mass personal communication, we area dealing with a connected lattice of interactions. Individual people become broadcast houses sharing their experiences with hundreds of others. With the advent of Twitter, we can expect our players to broadcast their gaming and gaming-service experiences in real time. To test this idea, we searched www.monitter.com (a Twitter-sponsored search engine of all things tweeted) for “Megabucks.” We found a series of discussions relating to recent jackpot wins, including many interesting opinions on the topic.
Now, if we take this a step further, we could have our casino employees monitoring Twitter content and keeping tabs on what players are saying about their casino, and in turn the casino employees could keep tweeters updated on their casino’s jackpots and service events. It may well be that tweeters would be more up-to-date with the current events happening at the property than the property itself.
This immediate information dispersal makes the communication mechanisms of the past far too slow. The implications are fascinating when you consider IGT’s ability, as the owner of the Megabucks jackpot brand, to join in the discussion with its customers.
By breaking the type of communication down into two dimensions, the first representing if the message is a dialogue or a monologue and the second representing if the message is delivered to a broad group of individuals, we end up with four groups (see below).
Targeted E-mail/Mail Custom Media
Broadcast Social Mass Media
Targeted Monologue (Mail)
A targeted monologue communication is normally conducted as an e-mail or piece of mail. In this dialogue, both parties are engaged in one-on-one communication. In marketing, this communication is focused on knowing the consumer’s preferences and suggesting personalized products and promotions to each consumer.3
One key constraint in this mode of communication is that when we design these promotions, we do not expect customers to communicate the details of their offers to other customers. This independence of offers enables marketers to establish a statistical control group, which they can use to measure the effect of the marketing program. This process enables the construction of A/B marketing programs that can determine the influence of one variable (for example did the amount of free play offered to potential players affect the amount of days the customers visited the property). A “simple t-test is the most common method of determining the results.”4 The following diagram shows the distribution of results that are drawn as a normal distribution and also shows how the control group is different from the treatment group.
Targeted Dialogue (Customer Survey)
With targeted dialogue communication, we expect a response to the request (the survey instrument). This kind of communication can take the form of a “survey,” either by phone interview or an Internet survey, to gain customer feedback. This feedback, while typically in much smaller volumes that the A/B marketing programs, has the potential of providing more information.
Broadcast Monologue (Media)
Mass media methods are used to broadcast a message to a broad and often untargeted audience. These methods are often associated with acquisition marketing efforts or general brand awareness efforts.
Broadcast Dialogue (Social Networking)
Last on the list of communication models is the broadcast dialogue and the social network. With the 350 million active Facebook users7 in December 2009 (up from 220 million in June 2009), we simply cannot avoid the fact that our customers now control a huge communication medium, a medium where they implicitly trust most of the information they share.
On Twitter’s website they describe how the network “connects you to your customers right now, in a way that was never before possible.”8 This immediate communication makes us a mere participant in the immediate dialogue that our customers are already engaged in. Now our customers can provide trusted advice on where to game and where to eat, sharing experiences while we think about measuring the response to the A/B marketing program for next month’s marketing efforts.
The Analytical Challenge
As this sea of change in communication sweeps the world, we need to address how we handle the analysis of our efforts with questions like 1) How can we establish independence of our control groups as our customers share information? 2) How can we determine the effect of a new gaming device when there is an invisible force of social networks influencing the games’ effectiveness? 3) Are we now forced to communicate with social networks and not individuals?
With the tremendous volume of real-time data, gaining insight into specific phrases that are being discussed is a first step in the analytical process. In the following example from www.tweetvolume.com, we look at the volume of communications. The results show the largest occurrence of the communication referring to the “Rio Casino” (see Diagram 2), which we find quite interesting. Combined with other information, like the current promotions or the conventions underway, we may find a relationship between marketing events and real-time social network chatter.
Now this communication becomes a measure of our promotional programs. For example, consider Graph 1 for an e-mail marketing program. The click-through rate shows the rate that people are responding to the e-mail. The Twitter references show the discussions referring to the property (we could also be more specific and have references to the promotion). This technique can be used to monitor the effect of events and gives some measure of the response to these events.
Gaming Floor Behavior as a Social Network
In our earlier series on market basket analysis, we introduced a series of techniques that enabled the classification of games into groups based on the products people tend to purchase together. The next section illustrates how the interaction of the customer with the gaming-floor results can be turned into a model that “sees” some of the influence of social networks.
In the following example we have found players who like to play next to each other at the same time. We have then applied an association rule analysis to find the players who have a lift effect on the surrounding players (i.e., those players who seem to have a positive influence on the number and nature of surrounding players).
The software AR Miner8 was used to find interesting rules with minimum support of 5 percent and minimum confidence of 10 percent and the results are shown in Table 1. Column A is the antecedent (or influencing play hour) and column C is the consequent or influenced play hour. A sample of these rules are shown below. One observation here is that players provide a lift effect on surrounding players within time bands; broadly speaking, afternoon players influence afternoon players and evening players influence evening players but there are no players who influence both.9
Social Networking’s Influence
With the almost staggering and world-wide adoption of social networking communication, it is critical that we find ways of plugging our companies into this stream of communication. Many companies are already being advised on how to handle the privacy issues with social networking10, and with the huge uptake in use, there is little doubt that social networking departments will become a critical component of how we manage the relationships with our customers going forward. To analyze the effectiveness of this new tool, a whole new range of analytical techniques will be brought in to play.
Communication with groups of customers can be measured by the number of people covered within one or two degrees of separation. Commonly, within social networking we look at how many immediate people are in our network and then how many are within one or two degrees of separation. Diagram 3 illustrates how almost anybody is connected within six degrees of separation.11
Social networking events can spread like wildfire—a bad customer experience broadcasted on Facebook can bring hundreds of people into the discussion, and Twitter messages can be retweeted immediately to spread information and influence in a viral manner. To counter this communication, a response will need to be rapid and will possibly escalate almost exponentially in order to handle the flurry of communication.
Given this huge river of social information flowing around our businesses, it is virtually impossible for humans to read and interpret the information flows. Enter mood measurements. These tools attempt to find the mood of communication. For example, if a significant positive social communication follows a new property opening it can be empirically measured.
Conclusion on Social Networking
Social networking creates an exciting new dynamic within the gaming floor. Your customers are now sharing information and experiences. Traditional A/B measurements, while still a powerful tool, are likely to be replaced with clustering techniques where we first establish groups of influence within our customer base.
The effects of this new technology are nearly impossible to predict, yet they are measureable. These measurements give us a chance to adapt the way we interact with our customers and become part of the social networking phenomena. The alternative is to close your eyes to these rapidly growing influences and hope that you make the right choices.
9 In future articles we plan to delve much deeper on how to apply the social networking association rules modeling on the gaming floor.
10 www.computerworld.com/s/article/9076678/ Planning_a_company_social_network_Don_t_forget_privacy_issues
11 The theory was first proposed in 1929 by the Hungarian writer Frigyes Karinthy in a short story titled “Chains.”