You’ve built a great online game, users are signing up at record speed and they are buying your virtual currency with the cute name – the actual cash is rolling in. Everything is great, your Board of Directors and stockholders love you! You’re in the middle of your first audit and the audit partner wants to talk about revenue recognition. The auditor starts asking questions about what the virtual currency is being used for; is it an item that can be used only once, does it deteriorate, or does it stay with the player throughout the life of the game. Then the auditor breaks the news to you, all of the revenue that has been reflected on your Profit and Loss Statement needs to be restated. You start to hear statements like “Enhancing the User Experience”. “What”, “but” … we made the currency non-refundable and non-transferable, isn’t that enough? The auditor says “No” and again talks about “Enhancing the User Experience”. So what do you do next?
Virtual currency is just that: virtual or fake ‘money’ used to buy items in online games & worlds like Farmville, Second Life and Mafia Wars. The game developer sells virtual currency in order to allow the users to purchase virtual goods (a sword, armor, or a plow for example) in the game they are playing. The user initially purchases a bank of virtual currency points, much like chips at a casino, and they are used as desired. More can be purchased, but they cannot be returned.
In analyzing the accounting technicalities of virtual currency there are generally three different revenue models that auditors are talking about:
Game Based – Under this scenario, the average life of the game would be determined and the revenue would be recognized over the remaining life of the game.
Item Based – Under this method, the revenue is recognized over the life of the virtual good that is purchased. Typically virtual goods fall into two categories, either consumable goods or durable goods.
User Based – Under this method, the revenue is recognized over the average life of the user.
So which method should you use? That is going to depend on the characteristics of your virtual goods and the quality of your data related to tracking the virtual goods. Let’s take a look at each of the three methods in more detail.
The Game Based revenue model will generally have the longest life and delay revenue recognition the longest. If your game has an expected life of three years, the revenue would be amortized over the three year period. Most likely your game will grow in popularity over the three year period as more users start to play. This method generally only makes sense if the other two revenue models are not feasible due to lack of quality data.
The Item Based revenue model is generally a good option if a large proportion of your virtual goods are consumable such as promoting your profile for 10 days or purchasing virtual flowers for someone that only live for one week. If the majority of your virtual goods stay with the player throughout the entire period that the game is played, then the average life of the durable item and the average user life will probably be the same. This method requires the ability to track each of the virtual goods that are sold and the average time it takes to consume the items.
The User Based revenue model bases revenue recognition on the average user life. As users come and go this will typically be a shorter timeframe than the average game life and result in higher revenue in the early years of the game. The biggest issue with this method is that you have to be able to calculate the average user life. This isn’t always an easy task.
How do you figure out the average life of the user? Initially the auditor suggests using the half life. How long before half of your users churn off? The problem is that half of your users never return after their first login. The half life of your user is less than a day. That doesn’t make sense. For one company, the solution was to use the 2nd half life, removing the impact of the first day effect. For instance let’s say we have 100,000 users sign up in month one. At the end of month one there were approximately 50,000 users still active. The company then looked at how much time elapsed before the active users were down to 25,000. This is the timeframe used for the average life of the user.
As you look at which revenue model makes the most sense for your company, the item based method should result in the fastest revenue recognition (due to the short shelf life of the virtual goods purchased), but is also the most complex to implement from a data mining perspective. The user based method falls somewhere in the middle, requiring more data than the game based method but is less data intense than the item based method. The game based method is the most conservative approach and requires the least amount of data mining. This method is the method of choice when data quality is an issue.
In the end, the method that makes the most sense for you depends on the characteristics of your virtual goods and the quality of your data. If you have a large portion of consumable items and good data quality, use the item based method. If you have mostly durable goods, use the user based model as it is easier to implement than the item based model. If your data quality is questionable, your only option is probably the game base model.
This is a new and evolving revenue source and many of the accounting firms have not yet commented on their stance. Like most new accounting issues, there will likely be significant debate and in the end the Big Four audit firms may disagree slightly on the methodology and the amount of diligence they accept for their respective clients. Ernst & Young published a white paper in April 2010 entitled "Revenue recognition on the sale of virtual goods." The most important thing that a company can do is document your data and have solid back-up support for the method used and how the timeframe was calculated.