One of the biggest questions on any slot floor is how to decide on the right level of participation games. This question of participation is a matter of huge debate and underpins an enormous rift between manufacturers and operators. One party might say that these games bring incremental revenue, while the other is questioning if participation games are just reallocating revenue that the casino would have collected. Let’s dig into this question and outline some real ways that, through customer preference and experimentation, we can discover the true value of a participation game. First let’s introduce ourselves to a couple of operator types when it comes to participation games.
The 2 Percenter Operator
The “2 percenter” operator believes that participation games should be less than 2 percent of the total gaming floor, and in many markets there are very successful operators where this low number is effective. Of course, games such as Wheel of Fortune are still a must, but the performance benchmark needed for the 2 percenters to introduce a participation game is extremely high. If the games are 20 percent revenue share and perform at double house average, then the cost of these games in participation fees is 0.8 percent of the overall gaming revenue.
The 10 Percenter Operator
The “10 percenter” operator has huge numbers of participation games that dominate the gaming floor and are a central part of the overall gaming strategy. An operator with 10 percent of its gaming floor showing participation games is paying 4 percent of its overall revenue in participation fees.
The True Cost
As we can see from the wide disparity between 2 percenters and 10 percenters, there is clearly much disagreement in the industry regarding the value of participation games, including wide-area progressives. The disagreement arises from the cost of operating one of these games. Most games have a fixed purchase cost (which can be paid either all at once or in daily increments), but wide-area progressive machines are priced based on a percentage of coin-in. Some of this percentage goes to increasing the progressive meter of the game (which is often more than a million dollars) and some goes back to the manufacturer, but none of it goes to the casino.
To understand how much more expensive a participation game can be, let’s look at an example. First, we’ll explore the cost as a percentage of revenue. The cost is calculated as a percentage of coin-in, but the true cost to the operator needs to be calculated as a percent of revenue.
To get started, let’s look at the overall cost of ownership for participation games compared to non-participation games. Let’s assume that the cost to purchase a non-participation game is $20,000 and that over the course of three years the game theme needs to be converted once at a cost of $3,000. Let’s also assume the game is winning $150 per day. So the cost to own the non-participation game is $23,000 and over a three-year period the game will give the casino $150 x 365 x 3 = $164,250 in gaming revenue, for an investment cost (as a percent of net revenues leaving aside cannibalization) of $23,000 / ($164,250 - $23,000) = 16%.
Now, for a participation game, we are still going to assume that it wins $150 per day, but we need some more assumptions. First, let’s assume the participation cost is 4 percent of coin-in and that the casino hold percentage of the game is 12 percent. Since the game does $150 per day in revenue, with a 12 percent hold it must do $150 / 12% = $1,250 per day in coin-in. Of this, the cost to own is 4 percent or $50 per day. So, over our three-year period, the cost to own the participation game is $54,750 with an investment cost (as a percent of net revenues leaving aside cannibalization) of 50 percent.
Our investment cost as a percent of net revenues has increased threefold with the participation game, and our gross expense has increased $31,750 over three years. Clearly participation games are very expensive, so operators need to know they are getting value for this expense.
Is the Cost Worth the Return?
Now we get to the controversy. Participation games tend to have average or better than average win per unit—and they are very popular with customers. But are they worth the increased cost to own? To measure this, we need to measure the cannibalization of each participation game. Whenever a slot machine provides a revenue lift (for example, replacing a $100 win per day game with a $300 win per day game), some of that lift is incremental and some of it is play that shifted from other games. The play that shifted from other games is called cannibalization. (For more information about cannibalization, see “Gaming Floors of the Future, Part 5” in the November 2011 issue of CEM).
So to understand if participation games are worth the increased cost to own, we will explore two simple examples, one where the game has low cannibalization and one where the game has high cannibalization.
For both of our examples, we will compare the value of a participation game that does $150 win per day versus a non-participation game that does $100 win per day. Thus, there is a lift of $50 per day driven by the participation game. We’ll also assume that the participation game has a 12 percent hold and costs 4 percent of coin-in. As we calculated previously, the cost to own participation game is $50 per day. In comparison, the three-year cost of $23,000 to own our example non-participation game is just $21 per day. Thus, in the examples below we want to see if we can justify the incremental cost of $29 per day to own the participation game.
The Low Cannibalization Game
First, let’s consider a participation game that is very popular and whose customers have shown a propensity not to gamble very much on other non-participation games. In this case, the cannibalization of this game from other games is low. If customers of this game don’t play many other games, there cannot be much play shifted from other games to this game. So let’s assume that the cannibalization factor is 20 percent—that is, of the $50 lifted from the participation game versus the non-participation game, only 20 percent (or $10) is taken from other slot machines on our floor. In this case, our truly incremental revenue from the participation game is $50 - $10 = $40, which is enough to cover the extra $29 in cost-to-own expenses.
The High Cannibalization Game
In this example, let’s consider a participation game that is also very popular; however, the customers who like this game also like to play other non-participation games. In this case, the cannibalization of this game from other games is high. Let’s assume that the cannibalization factor is 70 percent—that is, of the $50 lifted from the participation game versus the non-participation game, 70 percent (or $35) is taken from our other slot machines. In this case, our truly incremental revenue from the participation game is $50 - $35 = $15, which not nearly enough to cover the extra $29 in cost-to-own expenses. In other words, in this example, we actually lose money on our participation game, despite the fact that it is winning an extra $50 per day over our hypothetical non-participation game.
Notice that there are many factors involved in this calculation. Cannibalization is one of the key factors, but the floor win per unit is another major factor. The cost to own a non-participation game is fixed, whereas the cost to own a participation game is variable. Thus, for low win per unit per day floors, the increased cost of a participation game is less relative to the non-participation game, making it more likely that it is worth paying the participation fee. For higher win per unit per day floors, the cost to own a non-participation game is very small compared to a participation game, making it less likely that paying the increased expense will be profitable. Thus, market performance affects return on investment.
For example, in a low-revenue market, say $100 per machine per day, the simple economics of revenue sharing games can make a lot of sense. Quite simply, the revenue share may be less than a capital purchase. Daily fee games are a different matter, however, and in high-value markets, say $500 per machine per day, these games are more affordable as a percentage of total revenue. (See Table 1.)
Now let’s look closer at participation games’ cannibalization, as it is the major determining factor in our calculations of worth today. Unfortunately, trying to determine the effects of a new product on existing products can be daunting. Can we ever measure whether a specific customer’s $20 wager was meant for another machine or was in addition to his normal gaming spend on a “typical” trip?
First, let’s take a look at the player’s behavior using an unnamed but real customer database. When we attempt to determine the growth of a player’s worth, one metric we use is their ADT (average daily theo). This metric works fine when you are measuring trip worth over a period of time, for example, to see the effects of your marketing efforts. But if we use ADT as a baseline to measure a player’s incremental spend on a single day, the results could be disastrous. In a three-month sampling of data for the core customer base, we found that only 14 percent of all trips made by players were within a +/- 10 percent of their ADT. More than half of all trips made fell below the 90 percent mark of their ADT and 34 percent of all trips were above 110 percent of the player’s ADT. Nearly one-fifth of all trips didn’t even meet 25 percent of the player’s ADT.
As you can see in Chart 1, the variance in our player’s trip theo to his actual ADT is enormous, and we could easily miscalculate whether or not a $20 single session increased or decreased his trip’s theo in comparison to his ADT.
Ghostbusters and Market Basket Analysis
In our experience, most participation games can be shown to primarily cannibalize other participation games. For example, by performing a preference filter and cluster analysis, as described in the June 2012 issue of CEM, we showed that customers who played Ghostbusters in our unnamed database also had a high affinity for penny video slot participation products, both progressive and non-progressive, and penny reel slot progressive participation products. Customers’ activity levels in these game categories were high to very high, and no other product category had a high relative impact or activity level. (See Chart 2.)
The preference filter was an extremely important part of this. We limited the cluster analysis through the use of a preference filter to reduce the noise from the numerous trial sessions the product received immediately after its installation. Without the preference filter in place, this noise could incorrectly identify affinity products. For example, thousands of customers tried Ghostbusters in the first 30 days it was on the floor, most never to return to the game. This behavior is completely normal to almost any shiny newly installed game, so among the trial customers were the local video poker and video slot players. Without the preference filter, Ghostbusters players appeared to have an affinity for nearly every product due to the customers’ normal trial behavior.
Because Ghostbusters players showed an actual affinity for penny video slot participation products and penny reel slot progressive participation products games, we would expect that these categories would show the biggest effect of cannibalization, as the new games will take a share of the revenue within the affinity product category. Using a simple duplication of purchase model, we could assume that the share would be 2 percent since the new games would equal 2 percent of the unit count in the product category. Experience tells us that new games, particularly participation games, perform at least 30 percent higher for the first month or so following installation before they reach their stabilized performance numbers. It would not be surprising then to see that Ghostbusters has revenue numbers equaling more than 2 percent of our existing product category.
Now the question is, is the new Ghostbusters revenue incremental or did it simply cannibalize the product category?
Imagine we have a baseline revenue for the affinity product category of $2.5 million. The new product revenue is $115,000. We obviously can’t assume it is all incremental revenue, but how much is?
After comparing revenues before and after installation, we see that the existing product revenue shrank by about $100,000. From that, we can estimate the new Ghostbusters product cannibalized 4 percent of the existing product category, and we gained an incremental $15,000, for a 1 percent growth. (See Chart 3.) If the new product is significantly more expensive or less expensive than the product it is replacing, we would also need to look at the revenue numbers less fees to determine our net profit change. In this case, the new product was an addition to the floor, not a replacement, and the fees were in line with the product category average.
This may not always be the case, however. If your cluster analysis shows a high affinity to house games,when using the method explained above, you will have to estimate the cannibalization rate and determine if the new product revenue less fees is greater than the lost revenue to your house games category.
Wheel of Fortune to Move Players
In the Ghostbusters example, we attempted to estimate the incremental revenue and cannibalization of new product to the floor. In the following Wheel of Fortune example, we will show you how moving a bank of participation slots into a high-traffic but poorly performing location increased incremental revenue through impulse buys.
A bank of Wheel of Fortune games was originally placed in a difficult, poor performing area of the casino in an effort to lift performance in the area. It was against a wall in a small lobby area between the main floor and a restaurant but still near the main floor. Although the games performed well above average and the area average increased, we thought we could do better.
We wanted to place the games in a highly visible high-traffic area, where guests usually queued for the buffet and where there was plenty of foot traffic during entertainment events. Queuing guests and entertainment attendees were not playing the games they were waiting near and walking by, so despite being high traffic, the location was a poor performer. Chart 4 illustrates this move.
Formerly, Wheel of Fortune players were seeking out the game to play it. We were now putting it in full view of guests who were not seeking it out and were not intending to play it. When they did play, it was on impulse and during a time that they were not previously gaming (in line or going to or from the entertainment venue) and therefore did generate incremental revenue.
The entire move affected four banks, including one that was removed from the floor entirely. Even after removing a bank, we saw a revenue increase of 55 percent.
The increase in revenue and player counts happened immediately after the change, as shown in Charts 6 and 7.
But most interestingly, there was only a 25 percent crossover in the group that played these games in their original location and the group that played these games in their new location. The actual number of handle pulls was flat. And the average bet increased by 42 percent.
We have a case where we have more customers playing the game, but fewer sessions per customers because they are in line or passing by, and a much improved average bet—a much higher percent of max betters than before.
Are Participation Games Making Money?
Analysis of participation games is extremely difficult as we are tackling one of the most difficult analytical questions in gaming. The variation in strategy from the 2 percenters to the 10 percenters, combined with the market and the different pricing models makes for some real mathematical challenges. That said, as we have shown in our two examples, we can definitely change player behavior and, using preference filters, we can truly see the impact of these important games. As the industry continues to become more competitive, it is the authors’ view that these methods become central to good decision making on some of the toughest questions in gaming.
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. Ralph Thomas is Vice President of Strategic Analytics and Database Marketing for Seminole 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. He can be reached at ralph.thomas[at]stofgaming.com.
Jada Evans is the founder of MindSight Analytics and works with gaming companies to provide analytics and help develop strategies for a variety of topics including game performance, marketing, food & beverage, hotel & labor. She can be reached at jevans[at]mindsightanalytics.com.
Salinda Conklin is a seasoned slot operations and performance professional who has worked in properties in Iowa, Illinois, California and Nevada. Conklin brings a wealth of knowledge and over 20 years of experience to Silverton Casino Hotel as the Vice President of Slots. She can be reached at salinda.conklin[at]silvertoncasino.com.