April 26 - "Social Tsunami” Talking Games-As-a-Service with Playfish

Our April 2011 IGDA Montreal chapter event featured Kim Daniel Arthur, GM Mobile of Playfish, developer of wildly successful social games like Pet Society, Restaurant City, FIFA and Madden Superstars. About 95 intrepid devs brushed aside all concern for hockey playoffs to hear about the major shifts in working styles and thinking they might need to implement in order to transition toward delivering “Games As a Service” from a traditional product-based development background.

According to Arthur, Playfish was founded with the modest mission to “change how the world plays games,” with a focus on making play more about emotional and social interactions than interactions with the screen. Reaching 1,000,000 active users in their first six months online, they have seen 18 games go live on social networks such as Facebook since October 2007.

How exactly do you plan for that kind of growth? You don’t, explains Arthur. There will be mistakes; there will be surprises. Indeed, a closer look at the growth curve reveals several dramatic downward spikes along the way. Building a dynamic, diverse, and open-minded team with an across-the board commitment to user data analysis and a drive for learning and innovation is therefore imperative if you hope to thrive under the service model.

A tale of two business models
Arthur began with comparisons of Product and Service-based business models:

• In product-based development, high up-front production costs are met by a revenue spike at release, with returns dropping sharply soon thereafter. A risk-oriented approach that sees you betting on your product’s success, Arthur likens it to “having a baby and just sending it out into the world without raising it.”
• The service-based model sees users play, and as they do so, encounter opportunities to pay for premium features, over and over. In Arthur’s view, this is more sustainable and easier to invest in as costs are moderate and revenue grows over time.

Gaps in the Chain

• If a product begins its life in development, and is then passed to a publisher, who sells it to retailers, before it finally reaches the consumer, each additional step between the developer and consumer risks giving rise to a “false consumer” with needs and interests that need to be satisfied, but that might not directly relate to consumer needs. Revenue can also disappear into the gaps that occur at each step.
• A Service that is marketed directly to the consumer, with no third parties to impress, will have a much easier time responding directly to consumer needs, and to profit from the results. Under the service model, developers can respond to competitive threats by making small changes.

But this is perhaps more easily said than done. Under the Product model, players are motivated to try to like a game they’ve bought, because they’ve already paid for it  Not so, under a Service model, where the play itself has to convince players to part with their money.

Sales Funnels, Conversions & Longevity
Arthur describes the general Playfish user breakdown as 94% play-for-free, 5% paid, with just 1% of all users paying on a repeat basis. The recruitment of new players, the retention of veterans, and the conversion of both to paying players are therefore equally important.

Profitability hinges on three key factors:

• Distribution- Finding the players and going where they are.
• Engagement- Keeping them interested and playing- Playfish users average 30 minutes twice a day, but they’re working on extending that
•  Monetization- Figuring out what players will pay for and when.

Among the key lessons Playfish has learned, are that a compelling experience and a story that evolves over time are fundamental retention tools, and that every player action needs to contribute to the long-term gameplay.

Tracking the right data
Persuasive gameplay that encourages large numbers of new sign-ups while keeping veterans happy, and converting free players to paying fans is more than the result of simple intuition, experience, or guesswork. It comes from obsessive attention and responsiveness to data analysis— continually measuring and adapting to how people are using the game. “Everyone from artists to leaders needs to be trained to think about data.” Says Arthur. 

It’s important to consider a combination of quantitative data, qualitative data, and user feedback. But be warned that “the only true quantitative feedback is how people use the game.” Occasionally, players may gripe in the forums, but unless their actual gameplay reinforces their complaints, those should be taken with a grain of salt. For example, they may complain about the pricing of in-game items, but if they still want and buy the items, the point is moot.

Luckily, with more that 500 million analytics events occurring daily, there’s no shortage of user data for Playfish to mine for hints on how to adapt to meet user needs.

Understanding Your Own Point of View
Data in hand, you can set about solving problems and finding new opportunities— provided you understand the way you, as a manager, think.  “You have to understand your own point of view and how it may not be the best one for solving a particular problem,” says Arthur, suggesting you ask yourself  “are you a creative, non-cynical type, or can you be cynical?… I’ve learned that by nature, my own decisions won’t make very much money.”

Would you have made the decision to force the user to invite 10 new friends to play My Empire if he wants to complete the monument he’s been building? Would you have anticipated that in Pet Society, play-for-free gamers would eventually opt to pay a nanny bot to keep their Pets alive and happy while they’re on vacation, essentially paying not to play. These are the types of things many of us find irritating when we encounter them as players or observers, but they are the same mechanisms that feed the all-important 1% in above-mentioned funnel.

A constantly changing organization is a huge challenge
From a production standpoint, Games As a Service differ from Games As a Product mainly in that development continues post-launch— much like television, compared to film.  The update cycle for a Service model game looks something like:

Design>Dev>Deploy>QA>Deploy>Analyse>Live

With the launch as the beginning of the process, you have to be ready to staff up as soon as you go live. You have to have enough people working on the game post-launch to continually generate new content, assets, levels, and so on. Keeping your team motivated over the long term is a key challenge, as is keeping it organized and happy.

Arthur emphasizes the need for aligning cross-functional teams to minimize inter-group antagonism and keep handover chaos between development phases to a minimum. He suggests optimizing an organization around Service/Products, and creating vertically integrated teams. “Switching between independant departments creates a blame culture,” he says, describing one case where just having people sitting in separate rooms was enough to create a culture of “Server Guys vs. Client Guys.”

Games as a service can also carry a number of unexpected new staffing needs. For instance, once you have an economy in your game, you’ll also need pricing and merchandising people. Who else is going to figure out what’s going on when your milk-producing cow asset unexpectedly goes into over-production, and floods your economy with virtual milk?

Learn, Learn, Learn
The continuous nature of the Game As Service development cycle allows you to take risk because you can always tweak it later.
Allowing your team to make mistakes— even to fail— is important, because it’s the only way to learn from the data.

Arthur estimates that Playfish kills about half of all its development projects .

Launch Early
In the interest of learning now and tweaking later— or in showbiz terms, of always leaving them wanting more— you’ll need to get in the habit of launching early. This is guaranteed to be painful because there will be so many things you want to add or do better. And you will add them and do better. It will just be later.  Only experimentation will develop your sense of how many features you can remove before minimal becomes too minimal.

Distribution: Go where the players are
A fundamental decision Playfish made early on was to run their games on existing social networks such as Facebook, and the many-to-many distribution relationships they offer— a decision that paid off.  Indeed, even Facebook’s decision to take a 30% cut of in-game transactions hasn’t had as big an impact on the bottom line as expected because people’s enormous trust in Facebook also means that the number of paying users is growing.

Our event concluded with the usual applause, drinks, prizes, and the end of the hockey game broadcast on the screen behind the podium. Now that’s what I call “retention.”

See more photos on the IGDA Montreal Facebook Page.