Every day, millions of people play World of Tanks to clash with other tanks in virtual battlefields – and many of them have never paid a penny to play the game. Yet Wargaming, World of Tanks’ developer, has made billions in the gaming industry.
How did this gaming juggernaut monetize its free-to-play gaming structure and come to see the priceless value of every customer – even the ones who don’t pay a dime? In part, through its use of analytics.
The goal of Wargaming is to offer a great experience for players of all levels. Anyone can play for free on Mac, Windows, console or mobile versions, and players who want more can make in-game purchases.
Wargaming’s most played game, World of Tanks, has 110 million registered users online, and Wargaming collects data on every shot fired and each move made in every online game. “At any given point daily, we have about 4 million players playing our game,” says Alexander Ryabov, head of Wargaming Business Intelligence Data Services. “They play multiple battles, and those battles have multiple events in them that all generate close to several terabytes of data daily.”
Wargaming captures data from the second players log in to a game to when they log out. The company also collects and analyzes in-game chat logs, along with mentions of its games on social media sites and in many gaming discussion communities. Using this data, they can run models to retain customers, cross-sell other games, convert players into paid users, monitor the player journey and reduce friction points in the games.
In all, Wargaming processes more than 30 terabytes of data per month. It stores 98 percent of its data in Hadoop on an Oracle Big Data Appliance, with Cloudera managing the Hadoop implementation. Once the data is in Hadoop, ETL developers create data marts that integrate with SAS® to generate models and put them into production.
Scaling analytics to a terabyte a day
How can Wargaming take a complex model, and then quickly reuse that model's logic in thousands of other models?
Alex Ryabov, Wargaming Head of Business Intelligence Data Services, explains in this video.
Improving gameplay and customer offers with analytics
A team of data scientists at Wargaming develops models whose scores can be sent to an event-processing component in the game, to the company’s CRM systems and back to the team for additional modeling.
Recently, for example, the team recognized in the data that players kept dying in one particular place. “So they put up a hill in that place to balance the map,” explains Ryabov. “Our data scientists have created a heat map where you can see, on a game map, every shot fired during a certain period of time.”
The team also uses analytics to see if players are missing out on certain elements of the game so the game can send notifications for a better experience the next time. The message might tell a player where to access certain weapons or identify overlooked locations from a previous game.
“It will help players have a better experience in a game the next time,” says Ryabov. “This is just one example, but a lot of things like that can be accomplished using modeling and putting those models into production.”
To improve customer experiences even further, Wargaming applies text analytics to feedback collected on social media and in direct conversations with customers. “We can put certain filters in social media to get a sentiment analysis of overall play. We can also use sentiment analysis for customer support and to identify our all-star players on multiple channels,” says Ryabov.
How to operationalize large scale modeling efforts
Get expert advice on industrializing analytics from this TDWI Best Practices Report.
Get the free whitepaper
Scaling analytics to the massively multiplayer online experience
When Wargaming created its business intelligence program three years ago, it gravitated toward open source technologies. “Once we understood the need for in-depth data analysis and data mining, we started doing some initial, advanced analytics modeling in R, Spark, Python and all the other open source solutions,” says Ryabov.
But the team realized that scaling those initial efforts to thousands of models and more and more data every day posed big challenges. Describing Wargaming’s early use of open source analytics, Ryabov says, “The biggest issue for us was scalability. Our data scientists come up with a model concept, do some data wrangling, some data extraction and then we need to automate the results. It was all manual. It was a lot of work for our developers.”
According to Ryabov, the first models his team created took three to six months to implement. “Once we realized that we’re going to be running hundreds or even thousands of models for all of our games, all of our regions and in all of our time frames, we started looking for the solution that could make it scalable for us.”
After some in-depth research, Ryabov and his team found what they needed. “SAS Factory Miner and SAS Model Manager were perfect for our use cases,” he says, “because we can take the same model and multiply it by time frames, regions and by different products. So a model is virtually the same, but we can put it into the production environment, where we run, maintain and promote it over and over in an industrial sort of way. In our research, SAS was the only viable option.”
After the data is prepared and the modeling methodology is established, Ryabov says multiplying the model to thousands of similar models has become a one-person job. “To manually create and maintain that many models would take something like 10 to 20 people, and naturally they’ll make mistakes. An automated production environment like SAS does not make mistakes.”
The benefits of industrialized modeling
Automated and industrialized modeling has created many benefits for Wargaming:
- Transitioned most coding to a point-and-click based workflow for model building efficiencies.
- Reduced the amount of time needed to develop and deploy models by 60 percent.
- Reduced the need for data warehouse administration in the deployment and automation of models by 80 percent.
Overall, Wargaming data scientists are able to create and deploy more models in less time, which will result in higher revenues, better use of resources and lower opportunity costs. As the market grows and Wargaming continues to diversify into other platforms, it will be able to run even more models, retain more customers, acquire more customers and apply more complex analytics, all within the same analytics platform.
Most importantly, the players benefit too. “Our data scientists are a group of talented people that have very innovative ideas on how to offer players exactly what they want at the right time,” says Ryabov. “And SAS helps increase overall satisfaction and make the player experience even better.”
Plus, improving the gaming experience encourages more players to become long-term customers with a desire to invest in the game. “As our founder, Victor Kislyi, says, ‘Our goal ultimately is happy players.’ If players are happy, you know everything else will come.”
- Article A new arms race: Analytics for commodity market complianceRogue trading and dodgy deals are not the only things keeping chief risk officers awake. Today’s regulators now employ big data analytics to uncover troubles in the commodity swaps market. Staying ahead of innocent compliance errors – and quickly identifying the occasional bad actor from within – will require some tough analytics of your own.
- Using Hadoop as an Analytics CatalystLearn how to get the most out of Hadoop by developing a sound strategy for analyzing data in this new storage environment.
- Article How data visualization helps crack down on crimeAcross the world, law enforcement agencies are making more intensive use of data visualization technologies. They gain a real-time view of locations, layered with crime, traffic, geospatial, weather and other data. This means decisions are based on solid, robust data and resources allocated to guide intervention and crime prevention. Data visualization can make this possible for you too.
- Article How big data analytics can be the difference for law enforcementThe real value in big data analytics is that you don’t have to know what you’re looking for before you start. This is a huge benefit to law enforcement agencies that are facing tighter budgets and scarcer resources.