I’ve recently participated in an ongoing debate within the game industry about the value of “brands” to Free-to-play mobile games. My position on the subject is widely regarded as near heresy… which I completely understand, given how reasonable and widely accepted the view that driving traffic to a game is ALWAYS a good thing. As my regular readers know I recently developed an extremely powerful but complex mathematical model for game virality that I have dubbed the EUREKA formula. EUREKA is actually a set of six differential equations that completely model all possible first and second order properties of any online game. I devised and refined the formula to its present level of sophistication after many years of running MASSIVE online game networks and analyzing the monetization properties of thousands of games across millions of players.
I am fortunate to be one of the few people alive to have had access to such a phenomenal body of gaming data and to be in possession of the mathematical acumen to be able to construct advanced models that appear to accurately describe the complete life-cycle of all known online game dynamics. My motivation for undertaking this challenge was simple, it is extremely difficult to analyze the properties of online games because it is very hard to run properly controlled studies on a games behavior in the real world. It is nearly impossible to model social virality in gaming while keeping cohorts or samples of audience completely isolated from one another. I needed an accurate model to simulate game virality so that I could run virtual experiments on hypothetical games without needing to control a vast game publishing network, massive data warehouse’s and very sensitive experimental framework in order to gain insight into the fascinating and exotic monetization, consumption and marketing properties of online games. I believe that the EUREKA formula is a significant mathematical achievement but because of its complexity and the difficulties associated with testing it, it may be a long time before it is widely recognized as a powerful tool for building successful online games.
For the purpose of this discussion, I’m going to use the EUREKA formula in a highly simplified way to try to illustrate why artificially driving traffic to a Free-To-Play game will often have very little positive impact on the games overall life-time revenue potential.
To begin with, let’s define the widely believed assumption that we are going to examine:
Proposition: Buying or otherwise driving click traffic to a Free-To-Play game will generally increase its success.
Sounds obvious and reasonable doesn’t it? It’s certainly understandable that most people in the industry strongly believe this proposition. How would we go about testing this proposition scientifically in the real world? Ideally we would take a sample of random online Free-To-Play games. We would give them equal presentation to random audience samples and watch how they monetized with no marketing influence. These would be our “control” tests. Next we would buy an equal number of randomized traffic clicks to each game and measure their monetization over time in the presence of marketing promotion. We would then compare each games life time revenue from organic traffic acquisition versus the same game with pushed traffic to see which game performed better monetarily and observe any other interesting differences about how the games behaved under different marketing conditions. In practice, this experiment is very difficult to run because there is no simple way to get quality controlled audience samples that interact virally within the sample but not across samples at the same time. This is a great example of a situation in which a good mathematical model can help us visualize what might happen if we could run this experiment with rigorous scientific controls.
To conduct this experiment we’re going to setup the EUREKA formula in a simplified way so as not to confuse the discussion with an elaborate explanation on the meaning and interactions between the many parameters that comprise the EUREKA formula. A few simple assumptions can give us a clear NEW intuition for how Free-To-Play games actually behave without needing to rigorously tune the model parameters to capture all of the properties of an actual game.
The most current version of the EUREKA formula is publicly available on my Blog as a live Mathematica CDF file that anybody with the Mathematica web plugin can interact with.
Here are the simple steps to reproduce this experiment from the live EUREKA model.
1) Set the variable iExposed to 1. My online EUREKA model defaults to a sample potential audience of 1000. Setting iExposed to 1 means that of the entire available reachable online audience for a game (default 1000) only 1 person has initially been exposed to it.
2) Under “Marketing Paramaters” set Susceptible–>Exposed to 1. This is the number of clicks from the unexposed audience that you will buy for the game per day. In this example the game starts with 1 exposed player and adds 1 per day for 1000 days.
*Note that the default model automatically sets the Exposed–>Engaged parameter to .1 which means that on average 10% of the Exposed players will become paying Engaged players every day.
**Note that the default model automatically sets the Engaged–>Immune parameter to .01 which means that on average 1% of the Engaged paying players drops out of the game forever every day.
***$/DAU is automatically set to $1, again to give us a simplified relative view of how much revenue our hypothetical games will generate over time given different marketing and audience behavior parameters.
When we’re done the model should look like this;
The most important line on the graph is the red line. It shows the active Engaged paying players in the game on any given day. The blue line represents the unexposed players that have never heard of the game on any given day and the black line represents the players who have been exposed to the game and dropped out never to return. When the “Immune” audience reaches 1000, the games life cycle is over. Note that the game legend shows a figure for “Total Potential Revenue” of $58,753. This is the games total revenue generation at $1/DAU over the 1000 day sample period. Basically it is the area under the red curve times the $/DAU.
This graph represents a very simplified model of an online game. The game has ZERO ability to market itself. Players who play it have ZERO percent likelihood of recommending it to a friend. The only force driving traffic to this game is purchased clicks. To verify this, try setting the Susceptible–>Exposed Parameter to ZERO and the games monetization and traffic will vanish to nothing. Without artificial traffic pumped into it, this game won’t make a dime.
This scenario isn’t very likely however, even the worst games have some viral potential. To capture this, set the viral parameter Engaged to .1. This parameter simulates the idea that every paying Engaged player has a 10% chance of inviting another player to join on any day that they are Engaged. New players become “Exposed” and flow from Exposed to Engaged at a rate of 10%/Day in this scenario. We know that in reality not ALL exposed players will become Engaged players in a real-game which is what other parameters in the model are for but these details aren’t important to this demonstration.
In this scenario our virtual game with NO artificial traffic will organically generate life time revenue of $99,997, with peak audience engagement occurring roughly 175 days into the games life-cycle. Now we are set to conduct our virtual experiment in pushing artificial traffic to a Free-To-Play game. We simply take the model that includes some virality and push some traffic to it by changing the Market parameter Susceptible–>Exposed. Try setting it to 10 purchased clicks a day. What happens to the games life time revenue?
After buying 1000 days X 10 clicks/day = 10,000 clicks the game generates a total life time revenue of… $99,993… five dollars less revenue than the same game with no artificial traffic. What has changed however is that the games revenue peak moved forward in time to roughly 80 days. So according to The Saints BRILLIANT EUREKA model, a game that has traffic pushed to it can make less money than a game that has none… now either I must be an idiot who can’t model virality correctly… a distinct possibility… or there is something very deep and important to learn about how Free-To-Play games actually monetize. Now if you thought I was an idiot you probably wouldn’t read further so let’s proceed on the assumption that my math isn’t wildly out of whack. What happened?
If you look closely at the graph it’s kind of obvious. There is a limit to the size of every market for every game. EUREKA arbitrarily defaults to 1000 but the fact is that there are not an infinite number of potential players for every game. As players are exposed to a game and make a decision about whether or not to engage with it, they are consumed such that new available players become less common to recruit until the game has run its course and everybody who would possibly play it has eventually been exposed to it. When you push traffic to a Free-To-Play game all you do is accelerate the consumption of un-exposed audience which the game then churns through faster. If the game has any virality at all, virality is also amplified further accelerating the consumption of unexposed audience. But these parameters have NO impact on the games monetization parameters. If these parameters don’t change that game will always extract about the same amount of revenue from its potential audience. Pushing traffic at the game will just accelerate the rate at which the game will capture all of its potential revenue but NOT increase the games total revenue potential.
Since pushing clicks to a game costs money and virality is free, it always makes sense to optimize a Free-To-Play game for virality. A highly viral game doesn’t need traffic pushed to it to reach its monetary potential while a Free-To-Play game lacking virality will be very expensive to promote and make no MORE money than it would have had it been slightly viral.
If you actually want to make MORE money from a Free-To-Play game, marketing is not the way to do it. Try making small changes to the Engaged–>Immune parameter… better named the Churn Rate parameter and see what that does to your games Total Revenue Potential.
In summary, if you want to accelerate revenue you were going to make anyway from a GOOD Free-To-Play game, go ahead and burn cash driving clicks to it, just realize that you are probably not creating any significant found revenue. If you want to waste money promoting a BAD Free-To-Play game, drive traffic to a game that NOBODY wants to tell their friends to play, odds are that you’ll end up spending more on traffic than the game generated in revenue because the GOOD games in a network will drive up the cost of buying clicks above the value that a non-viral game will monetize at. Remember that a viral games generates compound $/DAU because in addition to monetizing it’s users it also recruits new users who will also pay to play the game. When it comes to the economics of Free-To-Play games our common sense intuition about how they will monetize is often WRONG, which is why I cooked up EUREKA to help everyone understand how to think about them correctly.