Will Work for XP: The Month a Million People Played My Game

In late July I wrote about the experience of demoing Dreamland for an audience of roughly 75 at the NY Gaming Meetup. As a quick reminder, Dreamland is a Facebook RPG with a central boardgame mechanic and a darkly cute art style, and it’s one of my main projects at FreshPlanet. We have an incredible team working on Dreamland, and there isn’t a day that goes by that I’m not impressed with the ingenuity and dedication of its individual members.

At the time of that writing, Dreamland had only recently entered into public beta and was pulling between 10 and 20 thousand monthly active users. This is a pittance compared to the juggernauts of the Facebook gaming ecosystem, but it’s a large enough sample size that you can start collecting reliable data on player behavior. After you’ve hit 10k MAU, your metrics are more shock-resistant: for example, a single player buying up $10 worth of in-game items in one day won’t drastically skew your Average Revenue Per Paying User (ARPPU).

All told, Dreamland’s metric-vitals were strong. We were consistently meeting or exceeding industry standards in terms of virality (measured as K-Factor), time spent per session, and sessions per user. Our updates with balancing revisions, bug fixes, and new features were showing positive results in player data. However, it felt as though we were holding our breath, waiting for our “Hockey Stick” moment: a period where a game experiences such explosive growth that the graph of it’s previous traffic is pushed flat as the scale expands dramatically.

On August 2nd, we hit that moment.

WTF?!

Considering the steady and predictable growth for the weeks after our open beta launched, you can imagine the jolt that went through the office when traffic skyrocketed from about 2500 daily unique visits to over 46000 daily uniques overnight. A special morning meeting was convened to explain where the traffic was coming from. Nice as they are, solid gameplay and beautiful art direction can’t account for such a deluge of new people.

As it turned out, Dreamland was chosen as one of a handful of games to be listed as one of Facebook’s “Featured Games.” The Featured Games section is located in a column on the right side of the Game Requests page — the place where game notifications go to collect digital dust when you don’t respond to them. Though I’d venture to say that most Facebook users aren’t aware of its existence, being included in one of these coveted slots can drive enormous amounts of traffic to a game. Also worth noting: since people will only see games they haven’t already played in this column, the traffic from this source is 100% new players.

In the month that we were a Facebook-approved featured game, Dreamland climbed to over a million MAU.

From 10 Thousand to 1 Million: Designing for a Sudden Influx of Players

Facebook doesn’t give anyone an estimate of how long they can expect to remain on the Featured Games list, and there isn’t an established precedent that we could look to. A game may be featured for a week or a year, and only Zynga games have made a permanent home on this prime piece of viral real-estate.

I’ve written in the past about my average day as a game designer. But when your player base begins taking off at something like 20 times it’s previous growth — a meteoric rise that may stop just as abruptly — “average” doesn’t cut it any more. The Dreamland team had no illusions; we all knew our days in the spotlight were numbered. We had to move quickly to capitalize on this massive body of new players.

That morning we drafted a new production schedule, one that was laser-focused on seizing the “low-hanging fruit” that would increase player virality and retention. Quick wins usually won’t generate big results on their own, but traffic in that quantity has a multiplicative effect on even the smallest positive changes. For the first week we were determined to maximize the effectiveness of our feed texts and images, the game’s logo, and our posts to the fan page. After a few days of testing we narrowed down options that that were getting the best reactions in terms of player sharing, click through, and community participation.

The real challenge began the next week. Our numbers had swelled, we’d tweaked and fine-tuned… and we were still on the Featured Games dashboard. There was little of that low-hanging fruit left to reach for, and committing to other features became a delicate balance. Now we wanted to make our updates punchy and worthwhile, something to convince Facebook to keep up our promotion.

Unsurprisingly, these big changes also tend to be the ones that swallow up the most development time. Almost all notable new features require intensive art production, game balancing, coding, and — of course — testing and debugging. Instead of racing to make the simplest improvements, we had to find the balance between two equally imposing forces: 1) do something cool enough to stay relevant, and 2) do it right freaking now.

To accommodate our new goal, we brainstormed a list of potential items to work on. Each had to answer to three main questions: Is it interesting enough? Can we do it in a week or less? Which metrics will it impact?

Here’s some of the ideas we came up with:

»   Add a new step to the tutorial to explain bedroom customization
»   Add more customization options in the bedroom
»   Prepare the game for language localization
»   Launch a limited version of “Candytopia,” the new world we’d been working on

Since some work had already been done towards it, we picked language localization. We also knew that Facebook encourages developers to implement localization so that it’s apps and games have a more global reach. Being able to report to Facebook that we were making Dreamland ready for folks in non-English-speaking countries would definitely give us some leverage.

The tactic worked: we were able to translate Dreamland into French as well as prepare the game to be translated into other languages, buying ourselves some more time on the Featured Games list.

The next couple of weeks followed the same pattern. It was a fast pace to work at, but definitely less feverish than the first few days. By the end of our run, we were able to introduce a couple new tutorial steps, add more customization options for players, and launch Candytopia. All the while our traffic numbers ballooned upwards, until they peaked above a million MAU and roughly 60k DAU.

The Afterglow and Readjusting

As expected, since we’ve been removed from the Featured Games section Dreamland has dropped to around 600k MAU. That number will continue to decline as time goes on, and will reflect something a little more “natural” in about three more weeks. It’s not hard to imagine that watching the traffic bleed off — artificially inflated or not — would be discouraging to the team. On the contrary, however, it’s been oddly reinvigorating.

For all the exposure the Featured Games list provides, it doesn’t come completely burden-free. The nice juicy carrot of viral traffic is without a doubt worth chasing, but the stick follows closely behind: under-perform and the well runs dry mighty quick.

In this regard, Dreamland was lucky. We already had plenty of nifty to-dos that fit into a near-weekly production cycle. If we’d been in a different stage of development, however, things could have been much rockier.

What if we hadn’t had art assets fully prepared for Candytopia? What if we hadn’t already begun packaging the game for localization? It’s likely that we would have felt pressured to create entirely new, less-anticipated features to impress Facebook, or to cut corners when implementing more elaborate features. That’s not to say that the work we did in that month didn’t greatly improve the game (it certainly did), but there is a hidden cost to developing within such tight constraints.

Once it was clear that the end of our featured game status was drawing nigh, the team got together for a special Friday lunch. Over burgers and shakes, we had the chance to really explore new ideas for the first time in a month. Compared to the careful analysis of only considering what could be accomplished in a week, the experience was nothing short of liberating. We’d been under a lot of pressure, and now we were getting the chance to relax and refocus our energies. For about the thousandth time since I joined FreshPlanet a little over a year ago, I felt honored to be working with such committed, creative professionals.

Ultimately, the benefits of being hosted on the Featured Games list are undeniable. Even though our traffic metrics are sure to take a big dive, there’s no doubt that we’ve come out ahead overall. Prior to the promotion we were averaging something like 2.5k daily active users, and we expect to level out around 22k daily active users post-promotion. That’s a permanent, lasting boost of nearly 9 times our previous DAU. All told, the estimated value of the promotion in advertising dollars is in the hundreds of thousands, and we didn’t pay for a cent of it.

Though I eagerly look forward to the day our games draw that kind of traffic again, I’m glad for the moment we’ve had to catch our breath. With the official launch of our newest game this week, however, it’s going to be back into overdrive soon enough!