Home Where’s The Money Now? Part 10 of 18:Theoretical Win Vs. the Tipping Point

Where’s The Money Now? Part 10 of 18:Theoretical Win Vs. the Tipping Point

Authors’ Note: In this article, we look at tipping points and video poker players, with a new twist on accurate individualized game data. The great challenge in the past has been that we know about players at a location and about the games that are played, but not about which games certain people play. This lack of knowledge has plagued the industry for years, but by applying an intelligent data-matching engine, we have figured it out. Remarkably, this new data set gives us a true calculation of theoretical win for each player.

In the March 2012 issue of CEM1, we wrote about the massive growth in data this century, and this massive growth in data is a profound new lens through which to see the world. This data includes information from sensors, social media and online gaming, and it is interactional in nature in contrast to the big data of last century, which included data from point of sale (POS), daily slot activity and carded sessions. Rough back of the envelope estimates show that big data is 1,000 times greater in size than transactional data, but the value comes from the connection of big data to transactional data.

Big Social Data and Tipping Points
This brings us to quite a different world of marketing today, where word of mouth online and “going viral” are forces that are both out of our direct control and enormously powerful. To understand this social effect, let’s apply some thinking from Malcom Gladwell’s “Tipping Point: How Little Things Can Make a Big Difference” (Back Bay Books, 2002). Gladwell describes a number of marketing programs under which a product “tipped” and became huge and how in this tipping process there are a number of key roles that need to be filled, including the Maven, the Connector and the Salesman.

According to Gladwell: “In a social epidemic, Mavens are data banks. They provide the message. Connectors are social glue: they spread it. But there are a select group of people Salesmen, with the skills to persuade us when we are unconvinced of what we are hearing, and they are as critical to the tipping of word of mouth epidemics as the other two groups.”2

Applying Gladwell’s roles in a gaming context, let’s examine three players: Charley, Mike and Sarah. (See Chart 1.)

Charley is the Connector. He is very active on social media but spends less than $50 per year in the casino.
Mike is the Maven. He is not very active on social media but spends more than $10,000 per year in the casino. Mike is, however, very knowledgeable about gaming, and his opinion is widely trusted. He does not make any effort to persuade people of his option; he is just a recognized expert.
Sarah is the Salesperson. She is a persuader. She is the one who can convince people that a new game or activity is important. She is not very widely connected, but she is very convincing. Furthermore, she knows and trusts Mike and his opinions. Sarah is an average-spend customer who spends about $1,000 per year in the casino.
If we base our marketing activities on a player’s spend per year, the majority of our marketing investment will be focused on Mike, with a mid-sized investment going to Sarah. We would largely ignore Charley. However, according to the Tipping Point strategy, we actually need all three players for a successful marketing initiative. If we target all three correctly, we may be able to bring about hugely effective initiatives that are like “marketing epidemics.”

This tantalizing and elusive prospect of making a dramatic impact with a marketing initiative is worth exploration—and we will explore it—but first we need to think again about big data. The huge media hype around big data is, according to Gartner, at the peak of its hype curve.3 At this point, the realization sets in for businesses that have come on board with big data that the process of gathering and monetizing this data is both difficult and transformational. Before we reach the tipping point for our marketing initiatives, let’s dig into one very important aspect of our data: statistical revenue or, as it is commonly known, theoretical win.

In the February 2013 issue of CEM, we described the fisherman’s analogy and how a fishing experience is in many ways like a gaming experience.4 In short:“If a fisherman catches one fish on average for every six hours of fishing, then the expected value of his total catch is 1/6 of a fish per hour. However, in reality … the fisherman either catches a fish or does not—there is no 1/6 fish. Furthermore, when a fisherman does catch a fish, he is likely to think he is on a run and will continue to fish longer. In other words, catching a fish (winning) affects the amount of hours spent fishing (gambling), which in turn affects the expected catch size (theoretical win).”

Given what seems to be a disconnect between the player experience and the theoretical win, why do we use it at all? The use of theoretical win is in fact very important, as it shows the expected statistical revenue (actually gross profit) to the casino, and for a large number of players, it is a very good way of understanding how much money you are making.5

Multi-Games Destroy Revenue Calculations
Of all the challenges to the theoretical win calculation, the outright error in using the average of the gaming machine is particularly mathematically destructive to the theoretical win calculation. To illustrate this, let’s look at a Game King game that has both keno and poker. The actual keno game has a hold percentage of 10 percent, while the poker has a hold percentage of 1 percent. The slot management system has the box set to a blended rate of 5 percent. Let’s go back to the Charley, Mike and Sarah.

As you can see in Chart 2, there is a sizable difference between the less accurate box theoretical revenue and the more accurate game theoretical revenue. Overall theoretical revenue reduced from $11,050 to $4,100, and at the customer level, the contrast is even greater. Instead of Mike being worth 10 times as much as Sarah, they in fact produce the same theoretical revenue. Now, clearly, before we start trying to make our marketing program go viral, having the correct data is essential.

The challenge with getting the correct data is that it simply does not exist in most deployed gaming systems. We opine that the reason that the player data is not matched with the detailed gaming data is twofold. First, the detailed gaming data is low quality. Second, the player tracking systems are quite separate from the slot accounting system. Given this near complete lack of precise player game level data and the clear need to have this data to complete accurate marketing, we have to dig into some complex computer science to provide the tools we need in the modern world of big data. The core complex tool that is needed is heuristic data matching.

Heuristic Data Matching 101
Heuristic data matching is an approximation whereby all the information in the data is used to determine who played what game. Think about matching the players Charley, Sarah and Mike to their individual game play. At first glance, it would seem impossible, as we only have daily summary values (most player tracking systems summarize data for each player at each game once a day). However, Mike is going to have much higher coin-in and lower actual win, and it is unusual for poker players (like Mike) to play keno (although it does happen). Sarah has quite different patterns. Given this knowledge, we could deduce that Mike plays poker and Sarah plays keno. This would be especially true if we looked at the data over a longer period of time. Heuristic data matching applies this “human” logic to match data and produce the relationship.

What is the best way to build these models? Well, we need to start with some basic assumptions, such as the ones described in the previous paragraphs (e.g., that poker players tend to have higher coin-in and lower actual win relative to their theoretical win). From there we can build a dataset of players who appear to be poker players and players who appear to be keno players. This data set will also include players who don’t clearly look to be one or the other, and those customers are removed from the set. Then we can build a predictive model that tags all customers of these games as either likely video poker or likely keno players. This model will likely include factors from our original assumptions on coin-in and actual versus theoretical win, but it will now also bring in details like what other boxes video poker players play versus keno players. For example, we may find that the keno players also prefer a certain group of video reel games with big bonuses.

In the end, heuristic data matching not only provides a set of data we can begin working with immediately, but it can also lead to more accurate predictive models that tell us with greater certainty which customers are playing which games on a multi-game box.

Even More Data Sources
Before we can really put our data to work, we also need to answer the question of whether the customer is a Connector, a Maven or a Salesperson. Certainly, to find out, we need to know our customers. Leveraging team members like players club reps, hosts and player development executives is a great start.

However, our customers also use our websites and mobile apps, in addition to third party social media sites. All of the data generated from these activities can be captured and appended to our players’ gaming data. With this data and the first step described in the previous paragraph, one can see a path that leads to a much better picture of our customers’ influence on our business.

Pulling It All Together
So there are two essential, and what seem like very different, parts to this puzzle. The first is that we have to understand our customers on a one-on-one level. This understanding is critical, and in the case of multi-game data, requires some special computer calculation methods and effort to bring the final results together. The second piece of the puzzle is that marketing to customers now requires that we understand different aspects of the customer’s non-gaming personality—basically we need to understand if the customer is a Connector, Mavin or Salesperson. Using this knowledge along with the player’s actual game preference enables us to execute some very special marketing programs.

To bring this down to the casino floor from the ivory towers of computer science and business theory, let’s illustrate how the combination of these two information sets can be used to adjust our marketing.

Charley the Connector
Charley is important, as he is a channel of communication. Using our knowledge that he loves poker, he is the perfect guy to spread the word about a video poker tournament. We could give Charley a bring-a-friend offer that was uncapped in terms of the number of friends but reasonably limited in terms of the size of the offer. Using this offer, Charley might be encouraged to connect with lots of people and bring them together to make for a successful event.

Mike the Maven
We know Mike’s opinion is critical, so we should make sure to ask him to help evaluate new game options and even consult with him about whether a game should be removed. It is crucial that Mike continues to endorse our products.

Sarah the Salesperson
Sarah gives us two options. She is the perfect person to convince others to attend the marketing event and make it successful, so we must make sure that she is invited and excited about it. We can try to grow our keno business via a keno event, or we can introduce Sarah to video poker and run a combined poker and keno event. (And if we choose to focus on just keno with Sarah, we can also look for another video poker Salesperson.)

Conclusion
Big data in the gaming business is no longer about simply tracking the customer play and rewarding them with offers based on theoretical win. We must now tackle numerous other issues, from the simple (yet surprisingly difficult) challenge of correctly tracking each customer’s theoretical win to the new and exciting challenge of creating social tipping points. But one thing is for sure: none of this can be accomplished without embracing big data.

Footnotes
1 See www.casinoenterprisemanagement.com/articles/march-2012/where%E2%80%99s-m….
2 Gladwell, M. (2002), The Tipping Point: How Little Things Can Make a Big Difference, New York: Back Bay Books.
3 See Networkworld.com’s overview of the report at www.networkworld.com/community/blog/big-data-reaching-peak-its-hype-gart…
4 See www.casinoenterprisemanagement.com/articles/february-2013/where%E2%80%99….
5 See www.casinoenterprisemanagement.com/articles/october-2009/who-due-back-part-iii-closer-look-theoretical-win.

Leave a Comment