This document discusses strategies for managing virtual economies within online games and applications. It covers topics such as controlling the circulation and scarcity of virtual currencies and items, testing prices and content, collecting user metrics, and using demographic data to improve monetization. The author advocates giving developers absolute control over their virtual economies in order to precisely test strategies, introduce scarcity, and maximize revenue through user transactions and purchases.
2. About Me CTO and SVP of Product at Live Gamer We’re a platform company: we provide virtual currencies, virtual items management, payment integrations, analytics etcetera to our customer base Basically, everything you need to run a virtual economy Over 70 customers– look on the website for details
3. Caveat I’m a CTO and you’re a potential customer But … I’m not selling you anything and this is not a sales pitch In fact, I’m going to use screenshots from companies that don’t use Live Gamer products Andy’s in the audience somewhere– feel free to talk to him
5. Recirculation In real economies, money circulates endlessly. “Multiplier effect” “Velocity of money” In virtual economies, you have sources and sinks
6. Absolute Control What can people own? How often can people buy things? How often can people trade? And with whom? Parental Controls Velocity Limiting Item Embargoes You can build systems that people cannot circumvent
7. Perfect Information Real world economic statistics are approximations “We can’t actually compute this, but we can kind-of-get-close by taking these other two numbers and dividing them” In the virtual world, the precision of your information is only limited by the size of your hard drive.
8. Physics and Scarcity Don’t Apply 1980: Lands End revolutionizes mail-order with their policy of “shipping before you could change your mind” 2009: Virtual goods ship in milliseconds What’s the incremental cost of another gif?
9. Testing is (Close to) Frictionless How does Walmart do price testing? Physically reprice all the inventory The cash register gets updated Happens at most once a day, is expensive, and doesn’t help a lot with demographics.
10. Discrimination is Perfectly Legal “I want to charge men more for ….” “I want to charge premium members less …” “I want to offer the blue coat for less IF you already own the yellow coat …”
11. Very Low COGS Sure, creative costs money And there are CDN costs for bandwidth But compared to real world items …. The incremental cost of another gif is zero
13. Rollout of New Content Staged accessibility Seasonality Level-based Constant creation of new inventory
14. Item Lifecycles New Depleted Light Use Heavy Use You control how things wear out (and whether they just go away)
15. Scarcity Rare items Only really matters As a call to action Or in a social setting Most effective with expressive items If you can’t interact with other players, scarcity is much less effective
16. Currency Paid and loyalty currencies Cash creation vectors How do people pay When are they prompted How do people earn the loyalty currency?
17. Velocity Limits / Embargos More about damage control in fraudulent circumstances than anything else “Don’t spend it all in one place” model for parent-child funding
19. Cash Out Vectors Can people take money back? What about player-to-player commerce? “If there’s no secondary demand for the items in your game, you probably don’t have a good game” -- Bethke
21. RPU “Revenue per user” Really should be “RFB” – Revenue From Bob Everything starts with knowing how much money you get from individual users And then rolling it up using demographic information
22. TPU Transactions Per User Transactions Involving Bob The most important measure of an economy’s health is: are transactions occurring. It’s also the right measure of engagement for applications that strongly incorporate virtual economies.
23. Velocity of Money Classically a macro-economic notion Micro-economic version Pretend the wallet is a FIFO queue. How long will it take the user to spend a coin you add right now? 147 days? That’s bad 12 hours? Pretty good
24. Available Cash Per User Analogous to savings metrics in real world Again, split it up by demo Equally interesting: split it up by recent actives Message users who have above average available cash and haven’t been back in a while
25. Stagnation Stagnant money is money that will never be spent It belongs to users who’ve abandoned the game It is in the “bottom” of wallets
26. Abandonment If your application is high churn And you’re not going to do much about it Then maximize the cash-in and don’t worry about item prices Separate out abandoned money in your metrics
29. Use Stores for Item Price Testing If you change the price on a single item , people buy something else. Instead, define and test price-sets and assign them to users Enables price testing with minimal side effects
30. Selling More Items If you can increase the number of purchases your users make Without lowering prices They will spend more virtual currency Which means they will need more virtual currency Which means they will buy more virtual currency Two basic strategies Constant introduction of new goods and services (new, newly unlocked, or rotated) Recommendations to drive additional sales
31. Recommendations Technology can be sexy Simple business logic is often best “If they look at the yellow coat, show them the blue coat as well” Offer bundles when users look at single items
32. Gateway Metrics What percentage of revenue do you pay to your payment processors? Do you know what payment processors your customers are using? What’s your fraud rate? What’s your abandonment rate in the purchase flow? Are you using the most efficient payment channels for your end-users?
33. Demographics Matter Do you know your users? Are you collecting information about what they’re doing?
35. Thank You Talk available at http://www.slideshare.net/wgrosso/managing-a-virtual-economy
Editor's Notes
And it’s very important to understand their behavior, and track them. Great article by Peter Thiel. One important point is he doesn’t believe we’re heading for inflation because lending institutions changed their recirculating behavior a little bit.
Zynga shows you stuff you can’t own!Offerpal blog: about converting between one and two currencies. In the real world, money is money. You can’t do this sort of thing. Food stamps try. But they just get sold at a discount on the black market
This is the core of the analytics argument. We track every wallet, and every transaction. We can compute new statistics that weren’t even contemplated. And we can macro-economic statistics from micro-economic data.
This is also a problem. Real world fraud systems are built on a slower scale.Modern credit card fraud is based on maximizing the number of purchases during a window of opportunity. It’s up to you to control things better.
And, of course, since this is a transaction in the computer science sense, it can only happen when the store is closed. We’ll talk more about this later, but it’s important.
A reviewer completely misunderstood this slide. The Robinson-Paxman act prohibits discrimination that I am not a lawyer
Shipping inventory to a store is simple. Yoville has 10K items
True for currency as well. Not enough people pay attention to rollover minutes and airline miles
ARPU spelled right!
E.g. please, forget clickstream analysis. It’s silly.
Odd thing; not sure what the point is“Breakage models”Driving the stagnation levels lower might increase abandonment. Might increase cash-in.
How much money is abandoned?
Stores.Think of it as Walmart versus Kmart versus Target versus NordstromExcept all you’re doing is changing all the prices.---- Multiple stores for a/b testing, for premium users, and so on.
Need image hereRemember: YoVille has 10K itemsStrategies are partially mutually exclusive
Recommendations as a toolRolling out new inventory strategy obviates collaborative filtering
I said “no cogs” up above, but that was about producing goods. The cash-in transaction has friction,hidden costs, and opportunity costs.
Can you store demographic information? If not, do you have some other way of categorizing users?Keep in mind that demographics is just “pre technology era clustering”Tracking PurchasesTracking accounts You get a CSR complaint The customer is pissed You give them some loyalty currency to make them go away Do you track why? Do you even know how much loyalty currency you gave away?