Background

Big Data is big in gaming. Take Zynga, the company behind FarmVille, Words with Friends and Zynga Poker. Zynga position themselves as makers of “social” games, which are generally played on social media platforms (rather than game consoles like Nintendo, Xbox or PlayStation) and take advantage of the connectivity with other users that those platforms offer. Their games are also built to take advantage of the Big Data those platforms enable them to collect. At their company’s peak, as many as two million players were playing their games at any point during the day and every second their servers processed 650 hands of Zynga Poker.

What Problem Is Big Data Helping To Solve?

Zynga have leveraged data to provide gamers (or bored office workers) with novel, compulsive distractions. And, of course, to make money.

How Is Big Data Used In Practice?

Zynga’s games and the hundreds of others that work on the same principle – for example the hugely popular Candy Crush Saga – use a business model which has become known as “freemium”. Players do not have to hand over cash up front to play them, although they often charge small amounts (micro-transactions) for enhancements that will give them an advantage over other players, or make the game more fun. For example, in FarmVille, which simulates running a farm, you can buy extra livestock for your virtual agricultural enterprise. Arrangements are also in place with a range of “partners” ranging from credit card companies to on-demand movie services, allowing players to earn credits to spend in the game by taking up their offers.

This ties into Zynga’s second revenue stream: advertising. While playing, you’ll periodically see adverts just like while watching TV or reading a magazine. Here, the data that they pull from Facebook is used to offer marketers a precise demographic target for their segmented online campaigns.

Big Data also plays a part in designing the games. Zynga’s smartest Big Data insight was to realize the importance of giving their users what they wanted, and, to this end, they monitored and recorded how their games were being played, using the data gained to tweak gameplay according to what was working well. For example, animals, which played mostly a background role in early versions, were made a more prominent part of later games when the data revealed how popular they were with gamers. In short, Zynga use data to understand what gamers like and don’t like about their games.

Game developers are more aware than ever of the huge amount of data that can be gained, when every joystick twitch can be analysed to provide feedback on how gamers play games and what they enjoy. Once a game has been released, this feedback can be analysed to find out if, for example, players are getting frustrated at a certain point, and a live update can be deployed to make it slightly easier. The idea is to provide the player with a challenge that remains entertaining without becoming annoying. Their ultimate aim is always to get players gaming for as long as possible – either to feel like they are getting value for money if it was a game they paid for or so that they can be served plenty of ads if it’s a free game.

Zynga make their data available to all employees, so they can see what has proved popular in games. So, even a FarmVille product manager can see the Poker data and see how many people have done a particular game action, for example. This transparency helps foster a data-driven culture and encourages data experimentation across the company. Indeed, Yuko Yamazaki, head of analytics at Zynga, tells me that the company are currently running over 1000 experiments on live products at the time of writing, continually testing features and personalizing game behaviours for their players. Zynga’s analytics team also do “data hackathons”, using their data and use cases, and they host many analytics and data meet-ups on-site. All this helps encourage innovation and strengthen the data-driven culture.

Elsewhere in the gaming industry, it has even been suggested that Microsoft’s $2.5 billion acquisition of Minecraft last year was because of the game’s integrated data mining capabilities, which Microsoft could use in other products. Minecraft, the extremely popular world-building game, is based around a huge database containing the thousands of individual items and objects that make up each world. By playing the game, the player is essentially manipulating that data to create their desired outcome in the game. Minecraft, in Microsoft’s opinion, provides an ideal introduction for children to the principles of structuring and manipulating digital data to build models that relate in some way to the real world.

What Were The Results?

Zynga measure success on two factors: internal adoption of systems and external player retention. Looking at the internal metric first, Zynga have 2000 employees, all of whom have access to the company’s data-visualization tool. At least 1000 employees are using the tool on a daily basis, demonstrating that the company have a really strong culture of data-based decision making. Externally, user numbers are around 20–25 million active daily users, which is a long way from their peak of 72 million active daily users in 2012. A number of factors are at play in this decline, including the end of Zynga’s special relationship with Facebook in 2012, and their historical focus on browser-based games (as opposed to mobile-based games). But, in 2014, Zynga acquired mobile specialists NaturalMotion, perhaps signalling a change of focus for the future.

“Compared to Web gaming,” Yamazaki explains, “mobile gaming has its own challenges, such as anonymous play activities, more genres of games and more concentrated session activities.” Particularly in mobile games, session length can be more important than the number of users, and longer sessions mean greater opportunities for Zynga. This is because in mobile sessions, players are usually paying attention the whole time during their sessions (whereas in a browser-based session, they may just have the page open on an inactive tab). So, though the number of daily active users is down, a stronger focus on mobile games will provide Zynga with the potential for greater reach and higher revenue.

What Data Was Used?

Zynga capture structured data on everything that happens in their games – almost every single play is tracked, amounting to around 30–50 billion rows of data a day.

What Are The Technical Details?

At the time of writing, Zynga are in the process of replacing their MemSQL database technology with MySQL SSD, running on Amazon Web Services. Their Vertica Data Warehouse is the world’s largest to run on Amazon.

In terms of future developments, the company are exploring real-time analytics and cloud analytics. Zynga have also started investing more in machine learning. And, in addition to the technology mentioned above, they now have a Hadoop/MapReduce environment for advanced machine-learning capabilities, focused on prediction, lookalike, social graph and clustering analytics.

Any Challenges That Had To Be Overcome?

Zynga’s marketing and sometimes intrusive presence on our social media screens has certainly come in for criticism, and it’s fair to say the company’s fortunes have declined in recent years – partly because of the ending of their close relationship with Facebook and partly because, in the tech world, there is always something new, shiny and often also free popping up to draw users elsewhere. The challenge for Zynga is more where to go from here, although rising mobile user numbers and new game launches offer some bright rays of hope.

What Are The Key Learning Points And Takeaways?

Zynga serve as a good example of a business built on innovative use of data from the ground up, and heralded the arrival of Big Data as a force for change in the gaming industry. Their culture of data-based decision making is admirable – something a lot of companies can learn from – and will hopefully stand them in good stead for the challenges ahead. As Yamazaki says: “Social gaming continues to evolve – from the way players play games to what features are available on devices … Zynga has killer infrastructure and insane data collection, holding records on billions of installs since its company launch. Big Data has always been Zynga’s Secret Sauce to launch it ahead of the competition, and will be a key to Zynga’s continued leadership in the space.”

REFERENCES AND FURTHER READING

Find out more about Zynga’s journey at:

1. https://www.zynga.com/blogs/engineering
2. http://venturebeat.com/2015/07/23/lessons-from-zynga-data-is-essential-but-it-shouldnt-rule-your-world/
3. http://www.it-director.com/enterprise/technology/content.php?cid=15336
4. http://www.forbes.com/sites/greatspeculations/2015/02/18/zynga-posts-disappointing-quarterly-results-and-future-guidance/