During our interviews for this book, what Gordon expressed was a common theme for those who had been through the design, development, and launch processes of a PW at least once. Business Models
Interestingly, most teams start design and development without a clear idea of their exact business model. This can lead to major confusion, with the pricing model changing several times during development, thus resulting in confusion among the team of just what they are supposed to be building and for whom. When you come right down to it, there are just two business models for online games: either the players pay or someone else does. Up to now, the three main models used for revenue generation have been advertising, promotions, and sponsorships; retail hybrids; and subscription fees. Advertising, Promotions, and SponsorshipsDuring the dot-com boom, which resembled a land rush with everyone trying to snare all the eyeballs possible in hopes of being able to find a way to monetize them later, the highly touted model was advertising. This was always a spurious model; well over 50% of all supposed "ad buys" on the Internet from the period of 1995 “2000 were simply ad exchanges between major sites, a sort of "I'll put up $10 million in value of your ads on my site, and you return the favor on yours." Each site then recorded a $10 million ad buy and $10 million in ad revenue. Naturally, someone forgot to tell the press, so the growth of Internet advertising looked quite nice in news stories. Internet users are notorious for not clicking through on ads. For a few years, before ad buyers woke up and realized they were being swindled, sites were paid solely by how many people viewed an ad. Now, ad buyers want to see evidence of click-through rates because that's a figure that means something to the advertiser. More and more, we're seeing payment based on click-through rather than the elusive and easily manipulated site visitors number, paid by CPM, or cost per thousand viewers . The ad-based model was never going to work for more than a few big sites anyway, and then only as relatively low incremental income. That doesn't mean you shouldn't do it; just don't think it will fully pay for development and management ”the chances are overwhelmingly against it. Promotions and sponsorships (in which an advertiser gets his/her logo plastered all over a web site or product for a certain period of time) are ancillary revenue generators that are seeing a small resurgence lately. This isn't creating much revenue for many sites; you're lucky to get $5,000 for a month of sponsorship or a promotion program. However, advertisers seem to see some value in it and it is on the rise again. These are all incremental revenue sources, however, and the amount of income you generate from them is never going to match expenditures; for that matter, online games sites are fortunate to have them make up 10% of costs. Retail HybridsSome publishers see hybrids as a good balancing point among the Golden Fleece-like rewards of intense player loyalty, visible Internet presence, a built-in market for follow-up products associated with a successful, actual online-only game, and the Scylla and Charybdis “like dangers known to be involved in actually developing, launching, and supporting one. The theory here is that Internet connectivity and multiplayer options in an otherwise solo-play home game boost sales. No publisher has yet provided proof of what those additional sales might be like, although common wisdom projects 10 “15% additional sales. To get a significant boost from adding Internet playability, however, you have to invest time and money into more than just adding to the retail game's code, as Blizzard did by setting up the no-charge battle.net host servers for their games. There is more than a casual connection between the fact that battle.net hosts hundreds of thousands of game sessions per week and the fact that Blizzard's Diablo II shipped two million units to retail establishments in the first publishing run and did a second run a couple weeks later of a reputed one million additional units. Compare this to the baseline of performance in the retail game industry, where an average number for first printings is less than 50,000 and first runs of even 250,000 are only seen for much-anticipated products from established developers and publishers. It is unknown what Blizzard pays in host hardware, bandwidth, and network operations costs, but the play obviously paid off for them. At the industry average calculation figure of $15 revenue per retail unit, Blizzard grossed at least $30 million in the first print run and something approaching $15 million in the second. It seems unlikely that Diablo II's development costs and the total operations costs of battle.net since day one added up to even 30% of that $45 million. The moral of this story: You'll get better results if you host the gamers yourself and host them well. Blizzard had a good game and good CS to offer. Just shipping with Internet capability might garner some extra sales; differentiating yourself by providing the whole experience will almost certainly have that effect. It is interesting to note that the casual evidence shows that the closer you get to the PW model, the better your chances for bringing in more revenue. Subscription FeesWhen it comes to making serious amounts of money with gaming online, the keys are PWs and subscription fees. This is where the money is currently being made. In 2001 alone, the world's top five subscription fee games had an estimated 2.8 million subscribers and brought in an estimated $207 million. With The Sims Online scheduled to launch in late 2002 and Star Wars Galaxies in early 2003, those figures are certain to escalate. Note the revenue figures in Table 2.1. While these revenue figures are small compared to the whole PC game market, compare them to the revenue figures a mere five years ago, in 1997, which were estimated at less than $40 million. That's a huge leap in only five years. If you add the myriad other smaller niche for-pay games into the total, 2001 was probably closer to $230 million. The industry is almost certain to top $300 million in gross revenue in 2002, and the figures here are conservative estimates. In other words, the potential of subscription-based games is enormous ”as are the up-front costs to make them, but the returns can be incredible. Table 2.1. Subscription Game Estimated Revenues , 2001 “2002All figures are estimates, based on public statements by the publishers or developers or experienced estimates by the authors. Note that Lineage 's revenues are heavily based on cyber caf fees from Korea, where most of the gameplay and subscription charges occurred in 2001. Lineage claims 4,000,000 registered accounts, but it is unknown how many of those are truly active, so a more conservative figure of 2,000,000 was used. Also note that for Dark Age of Camelot , launched in fall of 2001, the 2001 revenues are based on retail sales plus three months' estimated gross revenues, and 2002 revenues assume Dark Age of Camelot hit the 200,000-subscriber mark fairly early in the year.
Launch Costs
When trying to calculate the ROI potential of subscription-based games, you have to understand that the costs are much different from those of standard games. There is a huge disconnect in executives' understanding of the margins. The standalone packaged goods side of the industry sees margins of 60 “80%, while margins of 30 “40% for service industries are considered golden. Again, that seems obvious to the most casual observer, but it is also often overlooked. The industry gossip about why Take-Two dropped WolfPack's PW game Shadowbane from their development roster in early 2001 was they finally realized just how much it was going to cost simply to launch the game. [1]
Take-Two shouldn't feel lonely , because more than a few publishers have underesti mated or neglected to consider the costs of launching a PW game, too. Most publishers to date have made the mistake of simply keeping track of development costs, tossing in only some light figures for a couple of server clusters and a few people to answer player email, and letting it go at that. Until recently, it was extremely unusual for even $500,000 to be budgeted for a launch. Depending on how popular a game becomes during the public Beta test process, simply launching a game can cost $2.5 million easily, and costs of $5 million are not out of the realm of possibility. When you start adding up the costs for launching even 10 server clusters to handle about 30,000 simultaneous players (server clusters currently cost around $80,000 each), laying in the bandwidth, plus employing a 20- to 40-person player relations staff, 3 “5 community relations people, a live development team of between 5 and 20 developers, and a network operations staff to watch the hardware, pretty soon you're into serious money. The test process should give you a good indication of what your actual launch costs will be. In the meantime, at the start of this process, if you expect the game to reach 100,000 subscribers quickly, you'd be wise to budget at least $2 million in launch costs and be prepared to recalculate and scale up several times throughout the Beta test phase. It is difficult to give specific numbers because some costs such as bandwidth and salaries/ wages are variable by geographic region, as are usage levels and a host of other factors. Based on experience, however, we can give rule-of-thumb estimates on the costs for three months of expenses. Table 2.2 includes the two months directly prior to a game's launch, plus the month of the game's launch itself, for a game expected to host 30,000 simultaneous subscribers. Table 2.2. Launch CostsRule-of-thumb, three-month costs for a game to support 30,000 simultaneous players. These figures are somewhat low for a game expected to be very popular (100,000 “150,000 subscribers in the first three months). Bandwidth is not normally expressed as a percentage of costs, as the charges are normally calculated by multiplying the simultaneous player load by the sustained bit rate. However, our conversations with publishers indicated that this cost normally equaled about 20% of all the costs. Bear in mind that this can be made a variable cost that works in your favor if the team can keep the bit transfer rates from the server to the player and back low.
Note that the figures in Table 2.2 are rough estimates. The employee wages and salaries are monthly and include overhead, benefits, and tax loads. The bandwidth cost may be out of line for some regions , but was calculated at the current industry average of about 20% of total expenditures. The office space, PCs, furniture, and so on. field is an intentionally high one because there always seems to be something you forgot to buy. Also note that most PWs that launched in 2001 sold at least 30,000 retail units on the first day , so a load of 30,000 for pretty much any massively multiplayer online game (MMOG) should be the minimum you calculate. What That $10 Monthly Subscription Fee Pays for NowOne of the often- heard refrains from irate customers is "Spend some more of those millions on more CS and in-game development, you *&%$#&!" Believe it or not, most of the money paid for subscriptions already goes to these areas. First, there is the matter of how much money these games actually bring in. There is a mistaken belief among the customer base that you can just multiply the number of subscribers by $10 and voil , you now know how much money an MMOG brings in each month. In reality, an MMOG is fortunate if it can consistently, successfully bill 90% of its subscribers monthly. There are several reasons for this:
When you add all these up, a PW publisher is lucky to see 90% of the potential billables on a monthly basis. ExpensesOf that 90% that is actually billed and paid, what is that money spent on? Most people don't understand just how expensive these games are to develop, launch, and manage. This may surprise many players, but anywhere from 40 “60% of incoming revenue is spent just on keeping a game up and running. This means hardware, software, host sites, bandwidth (which is generally a variable cost that rises as data transmission totals rise, not the flat rate that many seem to believe it is), network operators to watch the game 24 hours a day, people to answer the phones and read and respond to emails, a crew of in-game CS representatives to handle problems and bugs , community relations people, a live development team to fix problems and add features and lands to the game it all adds up quickly. In a Wall Street Journal article in 2000, Kelly Flock, former CEO of Sony Online, noted that EQ used 100 employees to do all this at costs of $1.5 million per month. This would be slightly over 50% of the approximately $2.7 million the game probably recorded monthly at the time. After those expenses, there are the expenses of overhead and other development projects to add in, which can easily eat up another 30% of your $10. These expenses cover things like renting office space for all these folks, computers, desks, and chairs for all of them, office software, bandwidth for all their PCs, a local area network (LAN) to tie them all together, people to maintain the LAN, payroll and expenses for other development teams creating new PWs to create new revenue streams, a quality assurance (QA) department to test software, human resources people to make sure everyone in the company is paid and to track benefits, executive salaries, state and federal payroll and unemployment taxes, pencils, paper clips, stationery, tablets, power strips , heating costs in the winter and air conditioning costs in the summer, replacing the carpet when someone spills a pot of coffee, coffee, tea, hot chocolate, water coolers , copiers, fax machines, CD burners, white boards , conference room tables and chairs, telephones, telephone charges the list goes on. At this point, we've spent between 70% and 90% of the money paid in, leaving us with somewhere between 10% and 30% of the take. Where does that last bit go? Most of it goes to the parent corporation as contribution margin, which is then used to help pay overhead expenses there. State and federal governments tax anything left over from that. What is left over after that exercise is amusingly referred to as "profit." If you're very, very lucky (or your accountants are very good), that is 20%. It is probably closer to 5 “10% in the beginning, although you should be able to drive this to 30% at the 100,000-subscriber level. In fact, you really have no choice; if you can't drive it to 30%, chances are you won't make it. To improve margins, new subscription games began charging higher fees in 2001, most often $12.95 per month. Online gamers sucked it right up and kept on playing, as expected. Cost counts less when the player is getting something out of the game. If he/she is happy with the game, the price is fairly fungible up to around $20 per month or so. ROI Models for the NichesNaturally, each player niche has its own model for an ROI. Some players are just plain tough to get money out of.
This illustrates an important point to remember as you go forward with your analysis: The real money is in the subscription fees, not selling the retail package. At an average of $9.95 per month for access and trending higher toward $12.95 in 2001, someone who plays your PW for a year is going to pay you the equivalent of three to four retail games. The key is in not trying to gouge the buyer on the retail package, but to provide an incentive to the player to stick with you for years. Risk/Reward AnalysisPart of figuring out any ROI is determining the risk-to-reward ratio likely from approaching a particular market or niche within that market. Now that overly enthusiastic investors and developers have proven to themselves that just having a product on the Internet is not a business model (the old "just get eyeballs looking at our stuff; we'll figure out a way to make money off them later" ploy), it is time to start looking at the risks and potential rewards presented by the customer.
Understanding these risks makes it much more likely that you will create a profit margin. Properties to ExploitOne way to further improve margins is to reuse old code. Most companies just getting into this won't have core code that is of much use for a PW, but code from previous retail hybrids may be reusable for the next original project. Even for PWs, most developers have a tendency to ignore legacy code and try to build from scratch. There are some good reasons for this, the most important being that darn near every PW that has tried to use legacy code in the past has ended up costing more to develop than a from-scratch product due to reworking and refactoring the code. Training new people on old code doesn't necessarily make a lot of sense. However, there is a circumstance where it can work. In this, everyone should learn from Mythic Entertainment's successful example with Dark Age of Camelot ; some of the code in that game is well over five years old and many of the people who wrote the code worked on Camelot . The moral here is that if you've got the code and the people, use them. |