HBO Max Gets Personal

How do you separate yourself from the subscription video-on-demand competition? Real-time data analytics and content curation for nearly 70 million customers.

A generation ago, HBO encapsulated its ground-breaking brand ethos with a punchy motto: It’s not TV. It’s HBO.

That tagline distilled the inspiration that drove the network since the early 1970s when cable TV executive Charles Dolan launched a channel with an innovative approach to programming. HBO’s first show was mundane, a New York Rangers hockey game, but they began to create a lineup of shows that were uncut, commercial-free, and even R-rated. HBO attracted enormous ratings and subscriptions and built a stable of stellar programs — The Sopranos, Six Feet Under, The Wire, Deadwood, Def Comedy Jam, The Larry Sanders Show, among many others — that are cornerstones of the New Golden Age of Television.

HBO’s role as a destination channel for two generations of viewers stretches far beyond its ambitious programming. In May 2020, the network found a home in WarnerMedia’s HBO Max bundled alongside other iconic entertainment brands. This direct-to-consumer offering features 13,000 hours of premium content that includes HBO; Warner Bros. 100-year content collection; catalog titles from DC, CNN, TNT, TBS, truTV, Cartoon Network, Adult Swim, Rooster Teeth, and Looney Tunes Cartoons; classic films from TCM; and new Max Originals.

HBO Max launched when the subscription video-on-demand (SVOD) market was already getting mighty crowded. New services from Disney, Apple, and NBCUniversal have been released in the past few years, so HBO Max must find differentiating ways to connect with and keep SVOD customers. And they think they’ve found it: real-time analytics, content curation, and personalization for nearly 70 million HBO Max customers.

That’s where Duan Peng comes in. She’s the senior vice president who leads the group that applies the power of AI-based science and engineering to HBO Max, creatively wielding the power of the Databricks Lakehouse platform to create business intelligence on a massive scale in real-time. “Our job is to unlock the power of the data,” Peng said. “Understanding it helps us work smarter and more efficiently. It improves our decision making in every way.”

That enormous influx of data arrives in a steady stream from around the world. HBO Max has begun its global roll out and is now available in 46 countries after going live in Latin America and its first European countries last year. This data arrives in many forms, structured and unstructured, and is generated by streaming video and audio events as well as user interaction, text data, and subscription information. Merging these petabytes of information into a single system means insights develop faster.

The firehose of input is organized in a data lakehouse, a central repository that combines data from disparate sources and in a variety of formats. They then apply analytics and machine learning to the aggregated info generated by Databricks’ platform solution. The result is a unified platform where HBO’s data science teams work closely with content marketing, finance, privacy legal, and operations to extract real-time cross-functional analytics as well as broader business intelligence.

All this data flowing together allows Peng and her researchers to create a critical knowledge graph, employing Spark on Databricks to boost the performance of their analytic apps and quickly extract insights from streaming video events. A key goal is better personalization and individualized content recommendations based on customer viewing habits and other relevant information. Today, that comes from a daily batch process analysis, but the goal is to upgrade this to understand real-time behavior. For instance, a customer might finish a movie and then receive added viewing suggestions for specific actors, directors, or film genres. These arrive in the “More Like This” on-screen button. A “Click to Play” button measures suggestion effectiveness.

More data helps fine-tune recommendations for viewers looking for their next thing to watch, but what distinguishes HBO Max from its competitors is the power of human touch behind the curations. “We can analyze the metadata and how the consumer interacts with the content suggestion, including where they watch and what they watch – if our recommendations were effective,” Peng said. “We want the recommendation to come right away. This drives better engagement and customer satisfaction.” And, of course, higher loyalty and subscription numbers.

Increasingly, too, Peng is focused on natural language processing. HBO Max Recommends, for instance, is a voice-enabled experience for Amazon Alexa and the Google Assistant that helps the customers find something to watch quickly. If they’re interested in comedy shows they can say 'comedy series’ out loud and they get a recommendation from the HBO Max library sent to them directly.

It’s part of a deep and diverse digital transformation rooted in creating a natural experience, masking the sophistication of cross-media high-tech with the easy feel of a recommendation that seems like it’s coming from a friend. And it’s just the beginning. After all this,” Peng said, “we are still just starting. There’s a lot more we want to do.”

Learn how HBO Max sorts petabytes of info into business intelligence.

This story was produced by WIRED Brand Lab for Databricks.

Do Not Sell My Personal Information