At GDC Europe, GamersFirst COO/CTO Bjorn Book-Larsson discussed how the company rescued disaster-struck MMO APB and is taking it free-to-play, though not without some transitional pain.
The game is called APB Reloaded in its current incarnation -- under which it has bettered the pay-to-play predecessor game (which Book-Larsson calls APB 2010) by hundreds of thousands of players but is still in somewhat rocky shape.
Book-Larsson also serves as head of the company's studio division, Reloaded Productions, and revealed how -- with the help of a team, including some recruits from the defunct original developer of APB, Realtime Worlds -- "we took over this game for a few million dollars and how we continued to put more money into this game to make it free-to-play."
"The game died after less than two months. We were fortunate enough to be able to pick up the game and turn it into something new," he said.
He discussed how the game has changed in "significant ways" from its original release, but he also discussed how limitations of the original version have hampered the team's aim to change it as thoroughly as they would like.
On the original development at Realtime Worlds, said Book-Larsson, "The development process did spiral a bit out of control." He estimated that the original game cost $70 million to develop, with publicly reported figures as high as $105 million (he suspects some of that money went to Realtime Worlds' other, unreleased project, MyWorld.)
"To our benefit, the money really did build some incredible technology," said Book-Larsson, which includes custom tools for working with Unreal, extremely robust music creation and avatar customization tools for users, and a "great backend system to manage Unreal's crazy way to work with servers," he said. "For all of that money, enormous and cool pieces of technology were built."
However, the game still failed at retail. "What didn't that money buy? It didn't buy happy reviewers," he said, pointing out that APB 2010 has a 58 Metacritic.
"The biggest complaint about the game is that you just show up in the game you'd get killed and feel helpless... And the longer it was up, the worse it got. Really cool tech but really questionable design, particularly with how missions were handled," he said. "It removes player choice."
The game was only up for 52 days before it was yanked down again, so significant changes are necessary to make it appealing to players. "The long term solution was to make new maps and game modes that restore player choice," said Book-Larsson.
That is, he admitted, "extremely hard to fix... because the team has to work on building completely new player modes."
Three places where the team has implemented significant fixes include the skill rating system; the game auto-matchmakes players of similar skills, but the original system was inflexible. Since the original design forced players into missions without giving them any choice, and penalized them for quitting if they didn't like the mission -- or punished them with a loss if they were poorly matched or otherwise unengaged with this play -- the matchmaking became unreliable.
GamersFirst's solution was to "start over and create a proper skill rating system," he said. "The new skill rating system essentially puts you on a standard bell curve and it allows pretty good tracking of how you rate, even though you play in a multi person group. This kind of change has truly we think made the game better."
A less surmountable problem is the inflexibility of the RWTech map system in the game, a relic of the former developer's virtual world platform which disallows significant editing of level geometry.
This means that the game's maps "were hard to iterate on." For example, if you attempt to move a building "22 feet, you would have to throw away the map and start over," said Book-Larsson. This creates a problem in maps that have bad camping spots and choke points; the team can't do anything about them except move props.
Fortunately, "The maps aren't bad; they're just hard to refine," he said. All the same, "The solution to this is just leave it alone and build new maps that have less reliance on this [RWTech] component."
The game also had UI problems. Book-Larsson pointed out that players immediately dropped off after hitting level 10 -- "the decay of players after level 10 was astonishing."
It turns out that this is when the tutorial ended. "After you finished the tutorial at level 10 you didn't know what to do next," he said. "People would just run around in an endless loop and never realize they had to ready up to get a mission." Thanks to bad UI, in which players didn't realize they had to enter a readiness mode, many simply missed the core fundamental gameplay of the title entirely.
While APB 2010 hit its peak revenue on day one -- like many boxed product releases -- Book-Larsson said that it takes much longer to do so with an F2P game.
"Day 1000 is when a free-to-play game hits peak revenue. That's nearly four years," he said. "Games that can overcome the initial hurdles...and answer user requirements, there's no reason they can't stick around for a long time."
Conversion rates have been improving steadily, he said, and "we expect the long term trend to continue heading north." He pointed out that in APB Reloaded, 80 to 90 percent of the game content is available to free players.
But the game does sell guns in its in-game shop. To players, he said, "the big question everyone has -- guns, is that pay to win?" No, he said. "The team has done a really lot of work in trying to make the guns you buy...not overpowered."
"The design team has spent a tremendous amount of time trying to balance," he said.
The skill-based matchmaking system helps here, as it moves similarly successful players into competition -- and whether they're successful due to skill or due to buying new toys, it evens things out.
Speaking of buying items, 44 percent of items sold in the shop are weapons. "Premiums", such as convenience items, make up 35 percent, vehicles 16 percent, and player gear only 2 percent.
On August 4th, the company launched a reminder popup to get players to convert -- "and during that time we saw conversions go up tremendously," said Book-Larsson. Many players complained loudly about this reminder -- which pops up every 10 to 20 minutes of gameplay. "Did that mean we lost players who said they ragequit, 'I hate you all'? The answer is, no, we didn't."
"Do people actually do what they say that they will do on the forums? And if people are really angry with you will they stop spending money? The answer is no," said Book-Larsson. "Of course you can't piss them off" in significant ways, he cautioned, "But the smallest change will outrage the player who has been with you two weeks; they will feel like it's their game."
That's where metrics-based design comes in -- to separate feedback from actual player interaction.
"The biggest complainers were high on the biggest buyers." He said that ARPFP (average revenue per forum poster, he joked) "is about 30 percent higher than those who never engage with you."