Finding ALL The Money

[[Draft white paper, need feedback]]

Consider this blog entry a draft article on calculating a games optimal monetization properties using Quantum Economic theory.  Based on feedback I expect to iterate on this article over time until it becomes a high quality introductory white paper on the subject.  At some point in the future it would be fun to contrast the Quantum Economic pricing ideas discussed here directly to their Microeconomic counterparts but for the sake of brevity I’m going to plunge straight into the exotic world of HOW to calculate the optimal value of virtual products and design efficient business models to capture them.  To help me illustrate some of the more exotic ideas I’ve made some rudimentary slides that I also hope to improve on over time.

Let’s begin with a really simple equation for describing the “Value” of a virtual good or service.

Value = Potential Commerce Revenue + Potential Ad Revenue + Potential Viral Value – Marketing Expense

Or more succinctly;

V = C + A + V – M

For the purpose of this discussion we’re going to ignore Marketing Expense because calculating the optimal marketing expenditure for a game is a topic for another white paper of its own that relies on the result of calculating the other three critical variables in V.  The principal idea behind calculating V is to figure out what the MAXIMUM possible revenue may be achievable for a virtual good or service assuming that you had the PERFECT business model that captured all of that virtual products perceived value from the people who consume it at the point when they are most willing to pay.  In the classical world of product pricing this task is simpler and often easier to understand, but in the online world where you are trying to figure out a virtual products optimal monetization potential it can be much more complicated, in part, because many virtual products have NO PRODUCTIVE VALUE.  Games are of course the best example of virtual products that cannot have a rational value assigned to them using classical microeconomic pricing techniques.  Games have no use, they waste time and their perceived value is highly subjective to many environmental factors.

How much of this stuff are your games leaving on the table because you don’t know how to price them?

The purpose of figuring out a games “Maximum Value” is to more effectively measure how well a given pricing model, distribution model or marketing campaign is doing at capturing as much potential revenue as a game possess.  If a games “monetization space” is known then it is possible to know when you are reaching the limits of various approaches to maximizing revenue.  It also provides an effective way to measure how “good” a game really is.  For example, we all know that World of Warcraft is one of history’s most successful games of all time, however that does not mean that this great game was efficient at capturing all of its potential value, it may just mean that the game was so fantastic that it made lots of money despite squandering vast additional potential revenue.  What a shame it would be to have wasted billions of dollars in additional potential revenue from this game because it was not known how to calculate the games maximum potential value at the time that it was created?

In my own personal experience, developing these pricing techniques increased the revenue per user of my own online game company, WildTangent Inc. by 1800% per user over three years.

Let’s start slow and simple.  Value = The maximum potential revenue a game could earn over its lifetime assuming that it had a magic perfect business model and marketing strategy that captured every dollar that somebody was willing to pay for the content.  Value is just a number, however mathematically it is described by summing the area of a very complex multi-dimensional value space.  Visualize it as a giant undulating blob.

A crude attempt to illustrate a games potential monetization space with a single simple business model embedded in it, capturing a small area of the games total monetization potential

The purpose of this exercise is to develop a way to figure out a games total monetizable potential given all of the available monetization methods and devise more advanced business models that may combine a variety of approaches to efficiently capture more of a games maximum monetizable value.

Here I attempt to illustrate how different business models capture different subsets of a virtual products potential value. Classical economic thinking would have you conclude that you simply have to pick one model that captures the largest monetizable sub-region of the games potential. Overlapping models might confuse or anger consumers and result in cannibalization between approaches right?

WRONG! Good application of analytic heuristics to business model design can result in MUCH more revenue AND a better, simpler, consumer experience from the careful segmentation of audiences into monetization categories and the heuristic blending of models.

A few of the important dimensions describing this blob are as follows;

• The time consumption profile of the game
• The price/conversion sensitivity profile of the game
• The free play demand volume of the game
• The addiction or conversion probability profile
• The self-virality of the game
• The re-engagement value of the games players
• User purchasing power from their local and/or currency

It’s not necessary to understand all of these properties of the games monetization space to get a sufficient idea of how to estimate its volume or total value potential.  For our purposes it is sufficient to calculate 5 properties of this space to produce a good intuition for a virtual products total monetization potential.

Steps:

1. Calculate the games classical microeconomic demand curve.  In other words, its probability of converting to a purchase over a range of price points.
2. Calculate the games “addiction profile” which is basically a measure of the games probability of converting a user to a purchase at a negligible price at various stages in the game.
3. Calculate the games free play volume
4. Calculate the games negligible price play volume
5. Use the results above to estimate the games viral coefficient

We’ll cover the meaning of each of these calculations in more detail as we examine each of them.  The important idea to understand in advance is that there are three important variables we are attempting to assign a value to;  the maximum commerce value, the maximum ad revenue value and the viral value of the game.  Although virality is a strange and exotic property of virtual goods that can be very difficult to calculate usefully, we can use a crude linear approximation of virality to do a “sufficient” job of estimating the value space of most games.  My much more advanced articles on the EUREKA virality equations deal with this subject in much more depth and sophistication but for the purposes of this paper we will use a more classical linear idea of virality to come up with a reasonable estimate of a games total monetization potential.

For Step 1 we need to build a graph of the games probability of converting a user to a purchase over a range of price points.  This can be accomplished by offering the game to cohorts of test subjects at different fixed price points.  The result is usually a bell shaped demand curve.  This curve tells you the shape of the games demand distribution.  In a classical microeconomic world a marketer would use this information to assign a product the single fixed price that captures the largest area available under this demand curve.  In a Quantum Economic world however, this would be a mistake, because unlike the classical world of physical goods, online it IS possible to effectively offer the same product at different prices to different users based on their analytically estimated optimal price point.  Microcurrency based pricing is very effective at accomplishing this kind of dynamic pricing without consumers feeling that they are being treated unfairly by being charged different prices for the same game.

For Step 2 we need to figure out a games addiction profile.  In other words we are attempting to determine the point in the game play experience when the user has the maximum perceived value that they will ever have for the game experience.  Since this property varies from user to user, it is represented as a graph of user probabilities of converting to a purchase over a range of times or events in the game.  Since the goal of this step is to understand at what points in the game players have the highest perceived value for the content they are consuming, price is unimportant.  We need to conduct this test at the lowest possible commerce price in order to get the best quality data sample of purchase conversion rates across cohorts of users who are asked to make their first commerce transaction at different stages of gameplay.  In my experience a game that has its highest conversion value BEFORE the user plays it is a “bad” game.  A game that has its highest point of perceived value occur early in the game and sustain long into the game experience is a “great” game because it will convert users to purchase early in the game experience and retain them at a high monetization level.

Detecting a games “Peak Addiction” points is important because most forms of commerce transactions involve a high degree of friction related to the time and effort it takes to fill out a form and navigate the necessary security precautions that commerce transactions inherently require.  Commerce transaction points represent a major threat to a games audience retention and virality which in turn has a dramatic impact on the games overall value profile.  Knowing the best moments in a game to ask for money and having an analytical basis for calculating the probability you will get it is the secret sauce to engineering hybrid business models that use player behavior to determine the optimal monetization strategy for players on an individual basis instead of as a group.  If you know in advance, with a high degree of certainty, that a given player  WILL NEVER engage in a commerce transaction, how does it change the way you want to deal with the player?  A common choice is simply to discard the player, but almost all players have advertising monetization value and viral value in addition to potential commerce value.

In general, a good rule of thumb that a user is at or near their optimal point of commerce conversion is when they are self-returning to the game several times a week.  It has also been my experience that the optimal conversion opportunity is after dinner time during the week or best of all, around 2pm on a Saturday afternoon.  Asking a user for commerce during their lunch time can have a high failure rate because people often have limited time to enjoy themselves during lunch and want a quick entertainment fix.  Trying to convert them to a purchase during this period can risk losing a potential future transaction.

Knowing a games conversion profile can be extremely valuable because, in general, if a player fails to convert to a purchase during their maximum point of perceived value for the game… it is very unlikely that they will ever convert… which means they can be safely harvested for advertising revenue or viral marketing without fear of cannibalizing potential commerce revenue.  At WildTangent, we became 97% efficient at correctly guessing which game players would never pay us and converting them to ad revenue.

For Step 3, we need to calculate the “free” value of the game.  For games with classical expense structures like needing boxed retail distribution or having a high support cost per user, this calculation may be meaningless because the cost of supporting free players is not “negligible” for the game.  In a Quantum Economic world where serving and supporting online gameplay is relatively negligible, a games free value is very important to know because most games have extremely long tails of non-commerce monetizable gameplay.  To state it another way, if a game has a 5% conversion rate to commerce then it means that 95% of the games potential audience values the game at some price LESS than can be efficiently captured with a commerce transaction.  *In fact 40% of online gamers are often minors that simply cannot make online purchases for lack of a payment method.  If the cost of delivering and supporting the game is “negligible” then it is often a mistake to deny these users play when they simply cannot or will not ever pay for the experience.  Most games “long tail” of play demand, represents such a large volume of play, that if it were fully monetized via advertising it would generate as much or more revenue than the commerce revenue generated from the 5% of the audience who actually pays.

Consider a game that the average user plays 5 times a week for 20 weeks, resulting in 100 total sessions of play.  If 5% of the audience converts to a commerce transaction and has an LTV of $20, what is the advertising value of the 95% of the audience who is NOT paying?  Assuming that you could play a single $10CPM video ad in front of each free player per session you would generate nearly $1 in LTV from each free player, so a sample audience of 100 players would generate 95*$1 = $95 in ad revenue over their lifetime and 5*$20 = $100 in commerce revenue… almost the same money or double the money if you could capture both!  Clearly for many online games that have negligible operating costs, ad supported free play and commerce supported paid play can be very complimentary business models assuming they can be combined successfully.  As the addiction calculation step suggests, identifying people who will never pay is not that difficult and thus separating your audiences quickly into payers and non-payers makes it possible to capture both revenue opportunities efficiently.

Calculating a games free play properties is simple, make the game entirely free to a test cohort audience.  For many types of games, however, this is not as simple as it sounds.  For social games this can be difficult because the economic constraints of the games virtual goods market is part of the game design.  It can also be difficult for social games because, for the test to be decisive, you really need a player’s friends to all be included in the free cohort.  This challenge is a subject for another paper, suffice it to say for now that I have solved this problem sufficiently for social games in the past to get a meaningful read on the game by simply making the game entirely free in an isolated test market like New Zealand or Australia or by more sophisticated means requiring some modification to the games.

For this test, it is important to capture three dimensions of play behavior, starting with an initial test cohort of say 100 free players we need to learn the games retention rate over time  on a per play session basis.  For each play session the player engages in we need to record the date, time and duration of the session.  We also need to know how many friends free players interacted with during each session of play, how many they  invited to the game and how many accepted those invitations.

Step 4, calculate the games play volume at a “negligible price”.  Once we have measured how people consume and spread the game when it has NO cost, we need to compare that to the same data gathered from player behavior when the game has a “negligible” price.  “Negligible” in this case is defined as the lowest price you can transact in, usually around $1 in Western markets.  Basically we are measuring the basic transaction friction for the game and using it to estimate the overall adverse interaction that the games chosen commerce model has on audience retention and virality.

Step 5, calculate the games free and paid viral coefficients from the data gathered in steps 3 and 4.  The value of the games ability to market itself when it is free has to be included in the games Value equation for the maximum value calculation to be correct.  Every user the game can recruit for itself when it is free is a user that doesn’t have to be purchased or recruited via marketing expense for the game when it is paid for.  The property of online products to be able to do their own marketing is one of the fundamental properties of these products that is captured by what I have labeled Quantum Economic theory and is hugely influential to their online success but is not described usefully by classical microeconomic approaches.  As the EUREKA formulae described in my other blogs on this subject illustrate, it also takes some very exotic math to usefully describe the influence of virality on a virtual products success and overall potential value.  Here we will just consider a simple linear estimation of a games viral properties;  For our purposes we will simply take the ratio of the number of sessions played over two weeks by the free cohort audience to the number of sessions played during the same period by the paid cohort audience.  It’s not always possible to precisely track the viral acquisition of new players between test groups which is why making the game available in a controlled market free or paid and then comparing the normalized session behavior between market tests can be a reasonable alternative to precise viral tracking.  What we are really measuring here is the difference in play volume which happens to include the games viral properties.

Consider a game that attracts twice as many players for twice as many sessions when it is free compared to the same game that has minimal commerce requirements.  IF the game only supported a commerce business model it might “appear” to have a 5% conversion rate to purchase, however, if through the clever application of Quantum Economic business model design the same game was free to 100% of the players who would never pay for it the game would attract twice the audience… 5% of the additional free audience recruited by virtue of the free play mode would covert to purchase doubling the games commerce AND advertising revenue… producing 2X the commerce revenue as the paid only version of the game and 4X the ad revenue as the free game with NO virality would generate.  How much was the viral property of the game worth?  Roughly 500% of the games simple commerce value alone.  Using our simple viral equation the math would look like this.

Value = 1*C + 1*A + 4*V

Our ability to price our game correctly would be woefully inaccurate if we did not include the influence of virality on the equation.  For you mathematicians out there I know that this is an egregious simplification of the influence of virality on a games other revenue properties, but for the purposes of this paper I’m hoping to simply and strongly communicate the essential and fundamental influence of this property of online games in estimating and capturing their total potential value.  Including the influence of virality correctly in the math is at the heart of what makes Quantum Economics an important insight and analytical approach to pricing virtual products.  The important observation is that with good analytic tools Quantum Economics makes it possible to craft online business models that capture MOST of a virtual products potential value instead of just a small fraction of it.  It is NOT actually necessary, as it is in the classical business world, to choose one business model over another.  For virtual products, business models can be mixed, matched and blended together to maximize captured value.  My next blog will demonstrate how to use this data to construct more efficient hybrid business models that systematically extract more of a games potential value as a result of knowing how much value a game contains and where the value is located in the game.

Crude tests on hundreds of social games at Hi5.com and later Magi.com showed some significant improvement in retention and session duration as a result of replacing commerce offers with free ad offers

Overall monetization of social games doubled as result of the increased revenue/user from a mixed advertising + commerce business model combined with the amplified virality that resulted from increases availability of free play to non-paying players. The monetization yield increased over time as the compounding influence of increased virality and retention accumulated. Observe the “Quantum Economic” paradox that these games increased their commerce revenue by becoming apparently FREE to 97% of the audience… who was never going to pay anyway…

Comments

comments