I’ve written lots about the design of social games, explained the business model, and last week shared a free tool for designing virtual economies. This week I thought I’d write about what happens after the game has been built, how to figure out your marketing messages, and product testing on the cheap.
Having spent many of my teenager years playing Monkey Island I’ve become quite fond of pirates. So much so that when I’m designing a social game I like to imagine myself as a fearsome treasure hunting pirate. At this point you’re probably thinking I’ve been drinking too much grog – but bear with me while I stretch the pirate analogy thinner than Polly’s cracker.
Although I’ve never actually hunted for treasure, popular culture has taught me that treasure maps are not 100% accurate. It seems treasure maps indicate only approximately where the booty might be hidden, but through the collection of clues the resourceful pirate will find his prize.
Designing a social game is similar. At the start of the project we use a combination of research, experience, and prototyping to design the game. This design sets the general direction. Only when the game is live do we start to get clues (empirical usage data) that steers the design to a winning formula.
In other words, at the start of the project the game design is an informed hypothesis. But only by getting players into the game will we have actual data showing what they actually like, where they get stuck, or what to double down on. This is why you don’t want to spend too long trying to design and build the perfect game, it’s best to build a good game, go live, and start collecting data quickly.
This long rambling story about pirates is just to say that building a game is only the beginning. Next I’ll explain how to actually do all this data collection stuff without spending a pretty penny (last pirate reference, I promise.)
Low Cost Product Testing
While the goal of the usability testing is to make sure our target demographic can actually play the game – they know how to sign up, how to play the games, how to explore and interact, etc. However, usability focus groups are not the right environment for figuring out if the game is actually fun. In focus groups the child wants to do well, they want to please the moderator, and they’re more likely to give flattering feedback. Outside of a test lab that’s just not the case! In the real world players are bored, they’re looking for something fun to do, and they have short attention spans. They might give a game 30 seconds to a minute to see if its going to be fun, otherwise they’ll go somewhere else.
The purpose of product testing is to get a significant volume of real players into the product. We don’t use friends and family, or even beta users, because they won’t be representative of the real world. Using tracking codes in the game allows us to identify where the players get stuck, when they leave, what they were doing before they left, what they enjoy doing, and so on. This data driven process allows us to quickly analyse how thousands of people play the game, and what areas need enhancing.
Step 1: Create a scorecard
We start by creating a scorecard for the top 5 to 10 key metrics. The scorecard can then be used to answer the question, for any given range in time, for the 100% of people who hit our homepage, what percentage did X, Y, or Z?
For example, we might ask questions like, where do players get stuck, are they getting bored too soon, do they bounce of the home page, is the messaging right, do they remember to return, are they progressing too fast, are they completing the tutorial, how many are signing up, and so on.
Step 2: Bring on the players
The next step is to acquire a small, but statistically significant, volume of players. The easiest way is to run a small Google Adwords campaign – by bidding on the less popular key words we can be sure to find a low CPC so that we can drive 300 players per day for much less than $100.
Step 3: Let the data grow
With the scorecard, analytics, and Adwords campaign in place we can start gathering the data. After a couple of days we typically have enough data to start iterating on the product – changes to landing pages, tutorials, wording, and dialogue tweaks are all common. Because we have 300 players coming into the game every day, any changes are immediately tested, checked against the scorecard, and iterated.
This simple feedback loop has proven its worth time and again. It’s inspired by the military’s OODA Loop – Observe, Orientate, Decide, Act – and is really just a simplified version of that concept, applied specifically to creating a social game. Zynga use a similar process to iterate their games.
I recommend focusing on only 5 to 10 scorecard metrics because in a dynamic system anything that optimizes the sub-parts tends to sub-optimize the whole. In other words, we focus on total time through the loop, not on the time of any individual activity – we’re optimizing for the whole system not some particularly complex to collect metric. If we keep the loop simple, we can go through it faster, and optimize the game quicker.
Product iterations tend to last several weeks. The objective of this phase is to remove any obstacles that might be preventing users from completing the activities that drive the businesses Strategic Objectives.
Find Your Positioning
The product testing process also helps test our marketing messages. During product testing we’re buying Google Adwords in order to drive players into the game. These Adwords can tell us a lot about the product positioning and how to market the game.
Since we were only paying per click, it doesn’t cost us anything to cast a wide net. We can afford to run campaigns against every single product in an adjacent market space. We can quickly learn which competitive products are popular with our target market, and what messages get players to respond to our marketing.
The Brit Chicks game we developed is a good example of this process in action. Brit Chicks has elements of fashion, celebrity, dressing up, exploration, and friendship. All things that appeal to the 8 to 12 year old girl audience! We knew from the archetype research that different groups of players would be attracted to different aspects of the game. Using this information we created different advertising messages that targeted each archetype. The first step was to see which of these messages naturally start to drive clicks – which of the emphasized features or characteristics drive clicks.
Armed with the click through data we then created 3 landing page designs for each archetype – the Adwords adverts would click through to a customized landing page. For example, girls enjoy dressing up and we created an advertising campaign around dress up games, and a landing page promoting the avatar customization features of the site. From the targeted landing page the player has two options – register or leave.
While this process is initially run at a small scale and with very low advertising spend, in our experience the results scale up. The same positioning and same landing pages that convert at small scale can be used for bigger campaigns that drive more traffic. Designing and running the marketing campaign becomes cost effective, data driven, and very efficient
In summary, Google Adwords are your friend. Drive unbiased traffic cheaply, test your game, funnels, and marketing messages so when you’re ready to ramp up the marketing you can be confident you won’t be wasting your money. As John Wanamaker said “Half the money I spend on advertising is wasted; the trouble is I don’t know which half”. Only you do.