Home Where’s the Money Now, Part 7 of 18: The Human Side of Data

Where’s the Money Now, Part 7 of 18: The Human Side of Data

Authors’ Note: In this article, we dig into the gaming machine perspective that we, along with co-author A. K. Singh, described in The Math That Gaming Made1. We will also more deeply explore the concepts covered in Part 6 of this “Where is the Money Now?” series by looking at the human side of data. This human side of data might be messy to manage and analyze, as it involves intuition to decipher, but in many cases it can enable true value creation.

What is Big Data?

In the data world of the last century, big data was financial in nature, and a “grain” of data was typically one record per item(s) sold. For example, in retail, one grain of data was a single product, price and number of items sold. This century, data volumes are amplified to a level that makes the last century’s big data look tiny. This century, each grain of transaction data is surrounded by thousands of interactions. These interactions exist in World Wide Web tracking information—web logs—and in social media and extend all the way to remote sensing from satellite imagery. In addition, the grains of data themselves are no longer necessarily single numbers representing some aspect of a transaction. Often data comes to us unstructured (imagine, for example, trying to store every single webpage on the Internet every day), which has forced some data collectors to move away from relational databases and toward other forms of data storage, like highly parallel computer systems such as Apache Hadoop.

In the world of gaming, transaction data includes two main pieces: the gaming-operational data and the player-tracking data. This operational- and player-tracking information is related and has some major limitations. Here are two examples of this and their consequences.

Example 1: Multi-game/multi-denomination

On multi-game devices, multi-game data is not generally shown by player, and in some systems it is not even shown by machine. In other words, if a machine contains a $.01 version and a $.02 version of the same game, many slot systems will report that the machine won $300 yesterday but will not report that the $0.01 version won $250, while the $.02 version won $50. This is not only problematic for slot operators trying to optimize their slot floors but, far worse, marketers are often over- or under-valuing customers who play these multi-denomination/multi-game machines that are taking up an ever-increasing share of many slot floors.

Imagine a machine that has two games: Game A with a 10 percent hold and Game B with a 2 percent hold. Many player tracking systems won’t know which game the customer was playing, so they simply average the hold percentages of the two games and assume the machine holds 6 percent. What this means is that, no matter who is playing this game, the casino is guaranteed to be assigning the wrong theoretical win to the customer. A customer with $1,000 of coin-in on Game A should have a theoretical win of $100, but instead the system reports $60. Meanwhile, a customer with $1,000 of coin-in on Game B should have a theoretical win of $20, but again, the system reports $60 instead. In this example, the player-tracking system is going to cause the casino to either under-invest in this customer by 40 percent or over-invest by a whopping 300 percent. [Note: We authors are fascinated by the fact that these player tracking systems still dominate the industry. Imagine a slot floor with 70 percent multi-game/multi-denomination machines. Now imagine the slogan for one of these systems: “Correctly tracking your players 30 percent of the time!”]

Example 2: Session-based player tracking

Currently, most player tracking systems aggregate a customer’s play by slot machine by day. A significant amount of information is lost in this method of tracking. Imagine two customers, both of whom lost $60 over the course of an hour of play. The first customer put $60 into the machine and then slowly lost it over the course of the hour. The second customer had a much more volatile session, having lost $1,000 by minute 50, then hitting a jackpot to recoup most of that loss to finish down $60. These are obviously two very different sessions, but one wouldn’t know this looking at the player tracking data.

Fortunately, there are efforts to correct this in the industry. The current player tracking systems can be replaced with tools that track each and every wager that the player makes on a slot floor rather than summarizing by machine, solving both of the aforementioned issues simultaneously.

Figure 1: Transactions to Interactions2The graph in Figure 1 shows the exponential growth in data volumes and how we are now in the age of interactions.

Big data builds up in the most unexpected places, and sometimes its uses are not anticipated. There are several different kinds of big data, and some of this data can be combined with gaming data to promote insight-driven decision making. It may require a cultural change for many to accept data from outside of our sphere of control, but doing so is now critical to the management of casino operations.

This cultural change starts with an examination of the new sources of data. Big data today is often based around customer involvement with Internet-connected devices, so let’s dig in to some consumer- and consumption-based big data.

Consumer Consumption

Figure 2 shows the percentage of households that have adopted new technology over the last 105 years. These patterns are important to understand, as they put into context where big data is coming from and how it is pervading our society. Understanding this data may be the first step in making the cultural change necessary to appreciate the value of data beyond traditional transaction data.

Figure 2: Consumer Consumption3To see how the consumer consumption chart works, take a look at the black line labeled “Refrigerator” in Figure 2. Introduced in the 1920s, the refrigerator was only in 10 percent of households by 1930. By 1945, nearly 60 percent of households possessed one, and today the refrigerator is universal in U.S. households. Compare this to the microwave line in Figure 2. The microwave, introduced in the early 1970s, had already achieved 80 percent U.S. household penetration by 1990. So, while refrigerators currently enjoy a higher share of U.S. households than the microwave (100 percent vs. 90 percent), the microwave achieved 80 percent of households in only 18 years, compared to nearly 30 years for the refrigerator.

Following are four author-observed patterns in this data.

Pattern 1: Uptake is highly variable

The first trend is that the consumer uptake is highly variable. For example, compare the uptake curve of the radio line in Figure 2 compared to the lines for color TV and the computer. Even though these technologies span approximately 60 years, their uptake curves are very similar. Compare the dishwasher to the telephone; both of these curves, while they are 45 years apart, seem to follow a similar (slower) trend.

Pattern 2: Media technologies are adopted quickly

Looking at the radio, color TV, VCR and cell phone lines in Figure 2, we can see the adoption curve is less than 10 years. This trend indicates that new media technologies, such as online gaming and mobile gaming, could explode in market presence very quickly. However, this rapid adoption does not mean that older technologies do not get reinvented (see Pattern 3).

Pattern 3: Older technologies are reinvented

The relentless march of technology from 1900 to 2013 has shown a constant reinvention of technologies such as the telephone, automobile and washing machine, and these items have dramatically evolved from their first release.

Pattern 4: Modern Trends Produce Big Data

Modern consumer products produce data continuously. Compare this to the consumer trends of the 1900s when these early consumer adoption curves did not result in significant increases in data volume as they were used. Modern consumer trends all produce data as the consumer interacts with the devices on a daily (or sometimes minute-by-minute) basis. It is interesting to note that many traditionally non-connected devices, such as washing machines,4 are now becoming Internet connected.

Slot Machine Consumption

Figure 3: Real-Life Trending of Monthly Slot Machine PerformanceFigure 3 shows some real-life trending of monthly slot machine performance going back to January 2011. [Note: Any identifying features of the casino property have been stripped from this data.]

The y-axis in Figure 3 shows the relative performance (compared to the overall slot floor) of five games. Here, we see that the consumption patterns of specific products are very different than those of the consumption of technology. This illustrates how rapidly the picture of data can change when moving from the macro to the micro. More specifically, we also see that when making slot machine selections, customers behave very differently depending on the game—or, more likely, certain games attract different kinds of customers with different adoption curves and loyalty patterns.

Game 1 – This game was a raging success when introduced, and unlike Game 3, it was able to maintain a strong level of performance for the duration of this analysis, being above the floor average nearly every single month.

Game 2 – This game is simply a bread-and-butter staple of this particular casino. It’s always a hit, with no sign of slowing down.

Game 3 – This game started with big success, only to quickly decline in performance. In the end, it was removed from the floor entirely.

Game 4 – Never a strong performer, it has slowly declined over time. Now it is only played by a few devoted followers.

Game 5 – Unlike the other four examples, this game has a slow adoption curve but is now playing very strongly. After a poor start and a 6-month decline, the performance turned around in summer 2011 and has been getting better ever since.

The Big Data Challenge

Growth in data is happening at a rate that is transformational to the world, and this transformation will dramatically impact the gaming industry—the challenge is to keep pace.

The key to the big data challenge is in providing an analytical framework in which this data can facilitate insight-driven decision making. Tackling this analytical framework must involve data that is from beyond gaming, and this may seem daunting. To help keep it in perspective, there are at least two great first steps to take when tackling the analytics: analysis of space and analysis of time.

Land-based casinos are brick-and-mortar businesses that are full of spatial, or locational, data: for example, the location of every gaming device, the location of the restrooms and the location of the restaurants. Now while the virtual world of mobile devices is very flexible, much of the interaction data that is generated has a locational aspect. Think of this locational awareness as being generated by a combination of the GPS and Wi-Fi locational services. This common thread in the data is where locational intelligence enters onto the scene.

Locational data was once the domain of cartographers and geographers. Looking at Figure 2 we can see that many of these technology-adoption curves span many years. It is interesting to think that in the case of slower adoption curves, the technology has constantly reinvented itself over its adoption. This means that the technology adopted at the end of the curve is quite different to the technology that was in place at the beginning. Consider the automobile: it is hard to compare an auto from 1905 with one from 2005; the technology still performs a similar function, but its implementation is dramatically different.

Consider this example: A customer named Andrew is sitting at a gaming machine in one of his favorite casinos, Casino Y. He stops playing for a moment and uses his smartphone to search online for a place to have lunch. Andrew is creating GPS data that records his current location and the fact that he is probably hungry.

The restaurant app provider that Andrew used makes this data available to third parties (and generates marketing revenue), and these third parties use the data to place their marketing offers and events. Casino Z could be watching for restaurant searches like this and could be sending locationally targeted marketing events to Andrew. [Note: In future articles, we will examine technologies that are available today to execute these kinds of marketing initiatives.5]

Proximity to competition, or location, is the primary driver of a casino business, and this proximity data combined with census data provides accurate market-share numbers. Census data enables the analysis of the number of potential gamers by census area across a market catchment. What is fascinating is that spatial analysis also gives visibility to the catchment of competition and the effect of this competition on market share in terms of both visits and revenue. Managing this locational data and these spatial relationships requires locational-data storage, locational-query and locational-analysis tools. These locational tools are a basic building block of the big data analytical tool set.

Bringing it All Together

This review of big data shows how different big data is to regular data, as well as the cultural challenge it presents to a traditional operator. Like it or not, information about our customers is now generally available, and in many cases, our competition can today (or will in the near future)  analyze, market to and interact with our customer base. Organizational cultural change can be difficult, but there is no doubt that consumer trends are driving a new era of data. It is the operators’ choice as to when they will start to be consumers of these big data streams.


1 The Math That Gaming Made, Cardno, Singh and Thomas, 2013, Casino Services Publishing LLC.

2 “Where is The Money: Part 9 of 18”, Cardno and Thomas, March 2012, Casino Services Publishing LLC.

3 “How Americans Spend their Money”, extracted May 2013, http://www.nytimes.com/imagepages/2008/02/10/opinion/10op.graphic.ready…..

4 Refer to http://solarlight.metroblog.com/samsung_to_launch_internet_connected_was…, referenced May 2013.

5 Refer to http://www.casinoenterprisemanagement.com/articles/april-2012/where%E2%8… for more details.

Leave a Comment