How did EA DICE's free to play Battlefield Heroes take shape? Development Director Bjarne Rene delivered a comprehensive pre-mortem on what he would probably call the beginning of the project -- getting the game and the service that supports it up and running.
Battlefield Heroes is EA's first free-to-play project, and one of the first significant projects in the area developed by a major Western studio used to releasing packaged games -- EA's DICE studio, in Stockholm, Sweden.
Rene first identified the most important factor that shaped the game's development: "It all comes down to money -- Average Revenue Per User (ARPU). We look at something like Madden, a boxed product, $59.99 per year, per user. If you look at the only place [free-to-play games] are common, Korea, the numbers are a lot different." Users in Korea average three to six dollars per year -- so both EA and the team knew it must cut development costs. "That drove all the decisions we made on Heroes -- but I have to say we're not doing that good on the cutting costs part."
Coming from a traditional packaged game background meant that the developers had to change their thinking to succeed on this project. "We're giving the game away free, which presents some challenges. It must be cheap. We need to launch early, which puts a lot of pressure on getting something out there. And we need to integrate it with the web. It's an easy thing to underestimate, being a game studio."
Of course, it's not all about limitations -- it's also about opportunities. According to Rene, these include:
Short time to launch: "It's very cool to get the game built, and get it out there and see what people think and keep building it, rather than spending four years and putting something in a box."
Feedback, and player behavior: "We're better on the feedback than behavior tracking right now, but it's something we want to get to."
Building Battlefield Heroes
To reduce complexity of development, the team picked mature game tech that they were already familiar with: the Battlefield 2142 engine -- with "a few focused innovations." These include the new art style ("casual, friendly"), support for low system specs ("It's all flat and 2D-looking in a texture sense, but it ends up looking beautiful"), character customization ("That has driven a lot of what we've done in terms of making the game, including naming the game"), third-person perspective ("For casual players, we believe it's better to see you in the world" -- and you can see your character's purchasable stuff), and online distribution ("There's also the whole integrating it with the web and getting the microtransaction part done.")
The key of delivering a compelling downloadable game is not to cut features, but to focus on getting a complete game out quickly, limiting its scope in other ways. As Rene explains, "The base system is as meaty as a boxed product -- we can't just cut out subsystems. When we started out we had some pressure on, 'Make it a bit cheaper, make it quicker, knock it out,' but we found that we don't really know how to do that. And also that we don't want to do that, as DICE."
Rene maintains that the game has the same bar for quality as a boxed product. Though there was pressure at first, "It wasn't that hard to push back on it when the powers that be started to see the results."
The small initial scope of the game is the secret to the team's success. "We can't skimp on the features of the game, but we can go lower in that we only have two armies, two levels," he says. "We're trying to keep fairly limited [scope]. It's about the minimum we can get away with, but it's a fairly good base." For example, the game currently has only one mode, two teams, 22 weapons, and eight vehicles.
The dev cycle was short -- first production (an EA term roughly analogous to "first playable") was three months; production (proven and in progress) was four. "Importantly, for such a short development cycle, we had playtests from day one," says Rene. "The game mode we have now was pretty much done at the end of first production."
It's a Long Dev Cycle After All
Though the game will be released to the public after a relatively short cycle, Rene claims the game will enter "ongoing production. We're hoping that that phase is a number of years, because we'll keep on developing the game for as long as people keep wanting to play it. I'm hoping we're never done, but I hope that at some point we'll have a lot more stuff than a boxed product. It's pretty exciting to be allowed to keep adding to it."
Given all of this, the development team size shrunk as compared to a boxed product, but according to Rene, "Not as much as the people controlling the purse strings would like -- we still need full coverage."
To make it work, "We need multipurpose people," he says. That's not just about skill set, but also about attitude -- people who are willing to take on multiple related tasks, and not specialize. Developers tend not to think of online, continuous development projects like this as being as glamorous as working on a boxed product. Rene arguse, "Part of my job is to change the perception of it. I've heard this -- 'I don't want to be stuck on the live team.' Now if you tell me that I wouldn't be able to carry on once this thing is out there, I'd tell whoever said that, 'Over my dead body.'"
Moving Quickly with Scrum
Rene considers Scrum "ideal for this kind of project". In the project's backlog, "requests from everywhere can be captured... whether it's from the game team, from the community, a business thing... it can all go into the backlog, it all gets prioritized."
Though Rene says the team is not using Scrum completely, here's the processes they have adopted:
- Product backlog
- Sprint planning and review in two week cycles
- Daily Scrums
- Task level estimation
- Regular releases ("maybe every sprint, maybe every other sprint")
However, the team is not using Planning Poker. "I really, really like it," he says, but "we've decided not to use them because we don't really need that predictability."
Outsourcing for Success
Distributed development has been a major part of the project. The studios involved are:
- DICE: development)
- EA Shanghai: art outsourcing ("That's been working really well for us")
- EA Redwood Shores: online tech
- EA Guildford: QA, online development, legal, support functions
- EA Romania: testing
Putting it Online: Harder Than You Think
Rene has repeatedly encountered a cavalier attitude twaords the project. "How hard can it be?" they ask. "People know web stuff! It can't be that hard!" He responds, "Actually, it is."
His advice for those attempting a project like this, is to get people who understand the web. DICE hired a web development director, and a web producer. "Without those people, we would have never made it as far as we have," he says. He also recommends a web tech director, which DICE did not need to hire "because we had a team in DICE who were pretty strong."
The actual development of the Battlefield Heroes website was outsourced to a team of two to eight people, most of whom were on site at the DICE studio and sat with the game team during development. The web platform was built by EA, and is capable of supporting the game's business model, distributing and running the game, conducting microtransactions, and integrating advertising.
The site has social network aspects, because "we want the game experience to be sticky, and the whole experience is not just a fancy button on the net that says 'play now,' but it's a whole experience where you can see what you did, what your friends did, chat with people," says Rene.
EA's generic reference site was transformed into a very skinned Battlefield Heroes site. "Some components are very Heroes-specific, which we've gone and built ourselves, but for the most part we're taking what the platform team are building," Rene says. Testing is key: "It needs to be a lot more rigorous and comprehensive than the testing I've seen in the game industry." This includes functional, browser, integration, penetration, and performance testing. "The web people that we've hired have made a real difference in defining what we need, what we absolutely require." Testing must use the real site and real content "or it's worthless and artificial."
Running the Service
"It's kind of new to us and we haven't got it all worked out yet," Rene admits. After all, "There's a new game structure to support Heroes effectively being a new games business."
The larger EA organization traditionally handles all sales, marketing, distribution and support functions for boxed titles, but with Battlefield Heroes, Rene says, "Now that we're pushing the game out, [the dev team at DICE] want[s] to and need[s] to own more of these things." This means recruiting customer service, commerce, community managers, customer service analyst, a marketing manager, and ad sales manager.
One challenge has been getting the entire team into a service mindset. "Sub-team labels are dangerous," he warns. "Everyone's the Heroes team." It's been "essential" to be part of the same team, in the same room.
Rene boiled down his talk to a number of key lessons:
- Get the right people
- It's lots of hard work -- since it involves a whole new infrastructure for EA
- Most of the work is the web platform: "Do not underestimate it" by using existing game tech
- You're running a service from first public release: "We need to provide a real service to those [beta] guys. It's been real tough because we've been busy building the rest of it."
- The whole team is running the service; work on their perceptions: "Being part of the Heroes team, now that it's live, is really cool. You get to do stuff and two weeks later you find out what people think of it, rather than two years down the line."
Finally, the director notes that a healthy dose of paranoia helps the process. "It's kind of an in-joke but I think it translates -- we have to put question marks on everything," he says. "You've got to think, 'Can we do this right, can we do this better, does this make sense?' If I get a bad feeling about [something], I'll question it."