Dr. Dmitri Williams is a man with a vision in the games industry, who has for a decade and a half been practicing how to read ahead of the curve by analyzing past behaviors of users and players. Now, most analytics tools analyze hard data, like ARPU (average revenue per user), DAU (daily active users), conversion ratio (of non-paying users into paying users), and so forth, so that game developers and publishers can determine what attracts and keeps players in their games, hopefully as paying players.
Now, predictive analytics, according to Wikipedia, “encompasses a variety of techniques from statistics, modeling, machine learning, and data mining that analyze current and historical facts to make predictions about future, or otherwise unknown, events”. This leads us to Ninja Metrics, a newly minted and funded predictive analytics service, with the game industry as a large part of their focus.
If you’re attending GDC 2014, Dmitri Williams, CEO of Ninja Metrics, will be deep-diving into the analytics pool with not one but two talks. To get a head’s up about the secret sauce behind the cutting edge company he’s leading and the talks he’ll be giving, read on from here!
Events For Gamers: Dmitri, please share with us what steps led you from your research and writing work in metrics and data analysis, into the role as CEO of Ninja Metrics?
Dmitri Williams: I’ve been researching gamers for about 15 years now. When I started there were only a few of us doing this crazy thing and it was hard to even be taken seriously. My first career was as a professor, and I’m tenured at USC. While running a game research center, we started doing large-scale data analysis. It was “big data” 8 years ago—way before we’d ever heard of the term. Our team discovered some really powerful algorithms for predicting behavior as well as for predicting social interactions and effects. That lead to our Social Value idea, which had to be turned into a product and a company. I was the only one dumb enough to agree to run the new company!
E4G: Over the last few months Ninja Metrics has reached a high level of visibility among predictive analytics service companies, for the venture funding and unique service offering. Looking at the latter, what would you identify as the primary differentiating factors between the Katana Social Analytics Engine and the competition?
Dmitri: We offer all of the analytics everyone would expect, but we do two things that are unusual. First, we do predictive analytics—real Minority Report stuff about what players are going to do—quit, play, buy, etc. That’s difficult to do well, and we’ve turned it into software on top of that. The predictions are updated daily by a machine learning system that just goes on its own. So, no extra costs.
The more interesting we do is to calculate Social Value. This is the amount of play or spending that a player creates among his friends. We all have those friends who are fun and we like to play with more. Those people lead us to spend more time and probably to buy more stuff. Well, we’ve figured out exactly how to measure that playing time and extra spending. It’s been really exciting to see and it’s also very, very accurate. Our estimates are 92% correct over time. We find the “Social Whales” that developers want to acquire, protect and monetize.
E4G: Looking at the experience from an end user’s perspective, how might the game play experience be affected and shaped differently, by a company that uses Katana‘s predictive analytics functionality?
Dmitri: That’s going to depend on the developer. We give them the ability to treat players differently in terms of their game play, their offers, their communications, their customer service, etc. The most valuable players have traditionally been those who spent the most, but we’re showing how they are often the players who are leaders and the most fun to be around. Developers may elect to treat those social leaders better. How they do it will change game by game and different studios will have different philosophies about how to use the knowledge.
E4G: Generally speaking, how is social influence aggregated and leveraged similarly and differently between, say, players of MMO, web-based social and native mobile games?
Dmitri: Social value shows up in any activity involving humans, period. What’s different is that some activities are more social than others, and those ones tend to generate different kinds of relationships and interactions. Think about playing Solitaire vs. Tag vs. 5v5 basketball. Each one makes you interact differently and feel differently about your friends, teammates and opponents. Those rules and mechanics generate higher and lower amounts of social value. We find that the games that encourage good relationships generate the most social value. Consider an MMO where you need your teammates and you show up to help them and vice versa. Contrast that to a game where you play alone and post your score on Facebook. The latter is still social, but not as much as the MMO. The more truly social, the better.
E4G: What is the current status of Katana in the marketplace?
Dmitri: We’re live. Our production system is up and working, and is already supporting about a dozen developers. Some of the developers have massive PC titles while others have small mobile games. Katana is available to any developer, large or small and everybody gets everything. We price it based on the monthly active users so even small shops can get all of the tech at an affordable price.
E4G: What are some of the most interesting or unexpected patterns in player behavior have you learned through observation of the data during the beta test?
Dmitri: Some clients have been more interested in seeing Social Value by time rather than spending. It’s really cool to find players who genuinely create more play sessions. For some of these big Social Whales, they’re creating values 3, 5 or 10 times their own. For example, one player spent $100, but drove his friends to spend an extra $1,000. We didn’t know how big this could get and we see outliers like this all the time now.
E4G: What key trends do you see playing out in the future (over the next three years or so) of predictive analytics for games, and what role do you hope Ninja Metrics will play in being part of that future?
Dmitri: Predictive analytics are relatively new, and most developers don’t know how to get them. An automated service like ours makes it possible, so step one is just the education: you really can have this. Step 2 is teaching devs how to take advantage of it. Using predictive analytics can be used to proactively convert or retain players who are on the fence about a title. That’s going to drive revenue and devs will catch on fast. But predictives can also be used in AB testing. Imagine you tweak a game and see the churn probabilities go up or down. That’s pretty strong feedback for front-line developers.
Our part in this will be bringing it to everyone, and then spreading the best practices. By being able to see what works across our clients, we can advise them on the best ways to monetize and retain. We’ll also be able to share and show benchmarks.
E4G: Focusing for the moment on the GDC 2014 conference, can you tell us about the two talks you are headlining, Big Data Analytics 101: From DAU to Predictive Modeling and Social Whales: Understanding and Leveraging a New Kind of Player? What do you hope attendees will take away from both talks?
Dmitri: The Big Data 101 talk is the one I wish I’d had a few years back. I was bewildered by the KPIs. Heck, I had a PhD and could do advanced stats modeling, but I didn’t even know what KPI stood for. Now I want to walk people through the basics of analytics—what they do, how to get them started, how to use them, how to be smart about them, and what are the advanced techniques for power users.
The Social Whales talk will explain the idea of social impacts from basic psychology—why and how we impact each other. Then it will quickly move into the practical aspects of that for game developers—why you should care, how you can leverage it, and some thoughts about how it should shape both design and marketing.
E4G: Besides presenting two talks, what are the main objectives you hope to accomplish for Ninja Metrics at GDC?
Dmitri: We’ve invented this great new thing and most of the world doesn’t know what it is. So, goal number one is to spread the word that there’s a fundamentally new and powerful technology available. We want to get this into as many hands as possible. Otherwise, my goal will be to soak up information in the sessions, track the trends, and hopefully have time for a beer.
E4G: At which other game industry events might interested industry folks find you and Ninja Metrics over the next few months?
Dmitri: We were just at Casual Connect in Amsterdam, and we’re usually at E3, GDC Next (formerly ‘GDC Online’) and DICE. In the future we’ll be branching out to more GDCs as well as trade shows in other industries. We do love GDC, though, and here’s a special shout out to the Games User Research Summit during GDC where the usability and data folks come together to learn the latest. Coming from a research background, it’s always great to see the amazing things that people are doing in their labs at developers and universities.
– Paul Philleo, Contributing Editor
