Authors’ Note: Harrah’s started its customer loyalty program in 2002 with an investment of about $6.9 million; this figure grew to $95 million by 2005. 1,2 Free play has now become a major part of many marketing programs3, and depending on the system, it can be directed to specific players, games, areas, times of the day or days of the week. This ability to direct the behavior of players is a powerful instrument of change in managing the gaming floor. This article will first cover the math of free play analysis, and then begin to explore how this free play can be applied to downloadable games.
The analytical challenge of free play is that once it’s cycled through the machine, the money is like cash to the player. And that player is free to keep playing with it, move it to another slot machine, buy a sandwich with it, or take it home. When trying to analyze free play, some questions we therefore need to ask are:
1) How do we measure the difference in player behavior when they are spending free play, as opposed to the winnings of free play?
2) Is there a difference between the winnings of free play and actual cash?
3) How do we measure the effect of a return trip from players who are still ahead on their free play?
4) Should participation machines accept free play?
Accounting will normally treat free play as contra revenue. Simply put, this means that the revenue from free play is negative. In other words, actual win generated from free play dollars (winning back the free play) does not add to the total revenue of the property.
Building the Math
Step 1: Actual Win – Player vs. Machine
To build up the math, let’s start with a casino with one slot machine and one customer and look at the actual win of both that slot machine and that player (See Figure 1).
As was shown in our previous article on analytical perspectives, 4 different questions can lead to quite different answers, and this analytical perspective pattern continues as we examine the following questions. Three analytical perspectives here are:
1. Slot Analytics: How much actual win did this machine contribute? When looking at machine performance, we are interested in the total money generated by the machine so we can compare different games. Free play does not affect the actual win. (This becomes more apparent when we look at more than one machine.) The machine generated an actual win of 3,200, as the contra revenue does not get subtracted from the revenue of the gaming device.
2. Marketing Analytics: How much actual win did this player contribute? Here we treat the free play dollars as giving the player time on device; they do not contribute to the revenue from the player. The player contributed 2,000 in actual win.
3. Finance Analytics: Why did slots appear to make more money than marketing from the same player? Herein lies the problem. The challenge is how to allocate the free play costs to the slot machines so that the marketing and slots analytical reports reconcile.
Step 2: Actual Win – Two Slots, One Player
In this second step, we move to a property that generates the same overall numbers but that has two slot machines and one player. This second step shows how the free play can be allocated to the game (See Figure 2).
In Figure 3, the total height of the stacked bar is the actual win generated by the slot machine. The red shows the component of actual win from free play, and the blue shows the actual win minus this free play.
Now, going back to our analytical perspectives, we ask this question from the slot analyst’s perspective: Is Machine 1 performing better than Machine 2? Based on actual win minus free play, we might say that Machine 2 is outperforming Machine 1. From the finance perspective, we might in all likelihood say it generated less revenue, but to the slot analyst there is little doubt that Machine 2 is performing at almost twice the level of Machine 1.
An Allocation Model
To reconcile these differences, an allocation model based on the player’s actual win as a weight for the free play allocation gives a different picture. The actual win is used as a weight to spread the free play dollars spent across all games the player played on. The results are shown in Figure 4 as Adjusted Actual Win (win not including winnings from free play) and Allocated Free Play.
Now let’s examine the analytical perspectives. The slot analyst says that Machine 1 is a much stronger machine. The marketing analyst says that Machine 1 had the most marketing (free play) dollars allocated, but that the free play also had an impact on Machine 2.
The finance analyst is happy because—finally!—slots and marketing agree on how much money they are contributing to the property.
[Note: Using actual win as an allocation model introduces a random variation in the gaming into the distribution of costs. The use of other models, such as basing the allocation on the theoretical win, coin-in and Monte Carlo simulations of game play,5 makes even more robust models. We plan to explore these different allocation models in later articles.]
Consider two slot manufacturers, one who has games near the door and one who has games in the center of the floor. Both, naturally, are trying to claim the most revenue. Paying participation fees on free play is hard to justify to the operator at the best of times, but if the allocation model is used, we may see quite different customer behavior. This change in customer behavior includes players who wash their free play then move on to participation games. If they use the loosest games to do the wash, then the attempts to target free play have simply resulted in an unplanned change in customer behavior.
Adding in Tax
Taxation systems vary from the Michigan system of taxing free play as gaming revenue to the California system, which doesn’t tax it at all. When considering the actual revenue generated by the slot machines, taking out the effect of tax is easy enough when it is simply flat. However, once we allocate the effect of free play to the actual win, we can see what portion of our taxation contribution is being made from money we gave to the customers.
The Dynamic Gaming Floor
In researching and writing this article, we realized that to truly understand the dynamic gaming floor, the analyst must go beyond simple numbers such as actual win or theoretical win. Furthermore, once downloadable games and dynamic incentive programs having the right math models to understand what is really happening becomes the critical thinking, doing so may even be required to make these floors deliver on their capital investment. Simply put, to introduce downloadable games without the right math to model their performance is like flying blind into a storm. The good news for the gaming analyst is that few people are saying we can increase revenue by simply buying new gaming products.
Points to Ponder
Of course, this all gives rise to more questions, including:
1) What is the relationship between targeted free play and net win? Can targeted free play be used to increase revenue of a game?
2) Do people use looser games to wash their free play money?
Enter our mini casino analysis, where we consider different regions of the gaming floor and their relative performance—here, allocation becomes critical. If players tend to play their free play near the entry points, this will make the regions nearer to the door generate more revenue according to slots and less revenue according to marketing. At this point, before we even get to more esoteric measurements such as theoretical win, the debate rages—and debates over money can be the hardest to resolve. More questions:
1) Is there a difference between the winnings of free play and actual cash?
2) How do we measure the effect of a return trip from players who are still ahead on their free play?
Traditional analysis dictates that a customer who is given free play is playing “on the casino’s dime” until they pull money out of their own wallet. However, given the traditional metrics for analysis (coin-in, theo win, actual win, games played, etc.), measuring the point in time when customers actually pull money from their wallets can be difficult. What’s worse, even when customers do pull cash out of their wallets, we don’t know if the free play drove the customer to play more than they would have otherwise or less—i.e., did the free play drive incremental play or replace existing play?
First, let’s imagine that a customer, Bob, is given $100 in free play, and we want to know when Bob is no longer playing off this free play. Bob inserts his free play and begins to play. After 30 minutes have passed, we check with Bob and see that he has generated $500 in coin-in, $50 in theo win, and $25 in actual win for the casino (leaving Bob with $75). Is Bob playing off the casino’s money or his own money?
We can imagine both scenarios being possible. First, let’s assume that Bob has had a low volatility experience. After 10 minutes, he was up $5 (above his $100 in free play), then after 20 minutes he was down $5, and he finished down $25 from his original free play but still able to walk away with $75. Bob is happy, as he didn’t have to dip into his wallet once!
In a second scenario, Bob has a high volatility experience. Within the first five minutes, his $100 in free play is gone. He grabs $100 out of his wallet. Five minutes later, that is gone as well. Bob continues to dig and continues to have bad luck. After 25 minutes he puts his final $100 in the machine, for a total of $500 out of his wallet. Then he hits a hot streak and finishes his session up $75.
In both scenarios, the player has the same metrics for coin-in, theo and actual win, but he has two very different experiences, resulting in playing on the casino’s money in the first experience but not in the second.
Adding drop as a metric does not help this situation, since Bob could play his $100 in free play through once and break even. He could then cash out the $100, go eat lunch, come back and play the $100, and again break even. This could happen on multiple games, generating $600 in drop, as in Scenario 2, but resulting in Bob never pulling money out of his wallet.
The issue of free play on participation games seems, at first, to be a simple one. Free play generates coin-in, which in turns generates more revenue for the slot manufacturer (and less for the casino). Consider a simple example: If a customer is given $100 in free play and uses it on a participation game with a fee of 4 percent of coin-in, then the fee to be paid by the casino to the manufacturer is $100 x 4% = $4. That same $100 in free play has no additional costs if used in a non-participation game.
However, from the perspective of the slot manufacturer, the goal is to drive as much play as possible onto its participation games. Let’s take the example above but now assume that the $100 in free play lures the customer into playing $1,000 on their game of choice (including the free play). Assuming 10 percent hold, the customer would generate $10,000 in coin-in. The 4 percent participation fee would be $400, which exceeds the $100 in free play. The slot manufacturer actually has an incentive to offer free play valid on its participation games!
In today’s casino world this isn’t likely to happen, since the same manufacturer is making money through fixed sales and lease fees in addition to its participation games. However, let’s look forward to full server-based gaming. In our last article, we noted that server-based gaming may eventually go the way of the iPhone applications, where numerous software companies compete to develop the best games. Under this scenario, slot game providers would be under constant pressure to have the most profitable and most utilized games available. To bolster the performance of their own games, these providers may go down the road of offering free play incentives themselves, where the free play is only good on their product, rather than the casino offering these incentives.
Looking to other industries, we see examples of this. While grocery store chains offer their own discounts to their customers, there is a great deal of discounting done by the product manufacturers themselves, as they compete within a single store for the customers’ business.
In a future article, we will shed more light on some of the issues raised herein.
1 John Mills. “Auditing player reward programs.” www.thefreelibrary.com/Auditing+player+reward+programs%3a+casino+incenti... a0162353733
2 A.K. Singh and Andrew Cardno. “Who is Due Back? Pt. I.” Casino Enterprise Management, August 2009, pp. 10 – 16.
3 Andrew Klebanow (2009). “Examining the Value of Free Play.” Indian Gaming, pp. 82-83. www.indiangaming.com/istore/Apr09_Klebanow.pdf
4 A.K. Singh and Andrew Cardno. “Perspectives of Data: Financial vs. Slots vs. Marketing.” Casino Enterprise Management, June 2010.
5 A. K. Singh, Andrew Cardno and Anjan Gewali. “Why Do I Need Math? Pt. V: Skewness and the Player Experience.” Casino Enterprise Management, May 2010, pp. 18 - 23.
Andrew Cardno has more than 16 years of experience in business analytics, ranging from modeling health care 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. He can be reached at andrewcardno[at]yahoo.com.
Dr. A. K. Singh is a professor at UNLV. He has taught statistics, mathematics and operations research courses at New Mexico Tech and advanced statistics classes including Time Series Forecasting and Data Mining at Harrah Hotel College at UNLV. He has more than 80 publications in theoretical and applied statistics and can be reached at aksingh[at]unlv.nevada.edu.
Dr. Ralph Thomas is Vice President of Database Marketing for Seminole Hard Rock Gaming. During his years in the casino industry, Thomas has focused on maximizing profitability by applying statistical analysis to the company database. Previously, Thomas spent 15 years in academia, as both a student and a lecturer of mathematics.