How Twitch Is Using Analytics To Make Games More Fun

We talked with Kristin Chen, Senior Product Manager at Twitch, about the ways analytics is enhancing the game for users


Continuing our series of interviews regarding the future of gaming analytics and their increasing importance to the game in the lead up to our Gaming Analytics Summit this month in San Francisco, we spoke to Twitch's Kristin Chen.

Kristin is a Senior Product Manager at Twitch, focusing on Developer Experience and building Twitch’s Insights product so developers can optimize how they engage and invest in Twitch. Prior to Twitch, Kristin worked at LinkedIn on the Insights team focused on data-driven storytelling across the company’s B2B product suite.

Do you think analytics is driving gaming? Some people argue that analytics is taking the joy of out gaming. Is there a risk of that happening when it’s done badly?

I believe analytics are a huge driver, not only in gaming but in every industry. At Twitch, it’s crucial to better understand our customers and how they use our products. Data and analytics can reveal trends, the real-time impact of new features and, results of A/B testing at a scale that can't be captured from speaking to a small sample of players. From an esports fan and player's perspective, analytics can actually make gaming more enjoyable. Baseball fans, for example, track stats across players, teams, and leagues, while esports fans do the same when following the win rates of their favorite players. Players can also leverage sites and Twitch Extensions like OP.GG and Mobalytics that not only provide stats on professional gamers but also personal performance analytics. To prevent the risk of taking the joy out of gaming with analytics, we take great care in pairing quantitative findings with solid UXR research and we listen to our customers, which in our case are streamers, viewers, and players.

Does this allow you to better understand customer behavior in the gaming industry?

The Twitch Insights Team helps all developers understand how their games are performing on the service. We show Game Developers how our community of creators and viewers engage with their games and content via rich audience insights. These show the makeup of their streamer, viewer, and player audience on Twitch. We aim to provide metrics like stream language, location, and platform to help developers understand which audience segments are the most engaged or fastest-growing. If developers link accounts with Twitch, we get the option to provide more granular analytics about a player base on our platform. Our goal is to deliver actionable insights and recommendations to Game Developers so they can not only grow their Twitch community but also collaborate and integrate with Twitch. These insights will not only help Game Developers create games that are more enjoyable to viewers and broadcasters, but also enable Ecosystem Developers to make better tools. Our broadcasters will ultimately leverage these products to create better content and leverage data to better understand their audience.

How have advances in machine learning assisted with gaming analytics?

At Twitch, our Applied Machine Learning Team released the Deep Metadata API as part of the Twitch API last September. In the past, we have relied solely on metadata from game APIs, but we now have computer vision technology that extracts metadata through visual analysis of live broadcasts. By removing the need for streamers to link their accounts and 3rd party APIs, we are able to extend coverage to almost all live streams despite relying only on what is visible on the screen. Our long-term vision is to combine visually-scraped and game publisher metadata into a uniform set of Deep Metadata APIs so Twitch Developers can get all the video information from a single source. Currently, metadata on Hearthstone and Overwatch is available via the Twitch API. Ivana Savic, Senior Product Manager is focused on Twitch Machine Learning & Recommendations, and her team is actively working on adding more games, so keep an eye out for future updates!

Ok, so what is the most exciting use of gaming analytics right now for you?

What I’m most excited about is the data coming out of not only the watch and play engagement around a game on Twitch, but also viewer/broadcaster interactivity that arises from those games. We are facilitating those interactions and bringing them to the next level via Twitch Extensions and Twitch Drops viewer rewards. For example, Scavenger Studios developed a Twitch Extension for their new game, Darwin Project, that lets viewers influence live battles, like a talk show host influencing The Hunger Games. By giving players and viewers on Twitch yet another layer of interactivity with Extensions and Drops, we can provide a type of engagement data to developers not available elsewhere. Developers can leverage these analytics around interactivity to continue to enhance experiences for both creators and viewers on Twitch.

And finally, what do you see as the future of gaming analytics?

The future of gaming analytics lies in understanding the full picture of a fan’s engagement with Intellectual Property. Whether that’s when they are playing, streaming, watching, or chatting about it online; Twitch is the central source of that information. Our community is always inventing new ways to interact with games, video content, and each other, as seen in individual streams each day and the massive community-controlled gaming event that was 'Twitch Plays Pokémon.' We at Twitch are also inventing new ways for our community to interact, with Extensions and Drops being a direct result of that. With our dynamic community leading the future of gaming and social video, there is no doubt that the next evolution in gaming analytics and goldmine for player research lies in Twitch. In addition to interactivity, another exciting area of analytics is understanding how players learn and grow their skills in a game - how does a player rise to the top of a game leaderboard by watching pro players on Twitch?

To hear more from Kristin and other gaming analytics experts about how they are reshaping the industry with data insight, check out the upcoming Gaming Analytics Summit happening in San Francisco 18-19 April.


Read next:

Why We Need Data Visualization To Understand Unstructured Data