How AMC uses machine learning to find out more about TV viewers

By | 4 minute read | January 3, 2019

Machine learning is a hot topic no matter the industry, and rightfully so. Many see it as a path to greater efficiency and deeper insights.

This is particularly true for television, where rapid industry changes are evident to even the most casual observer. AMC Networks, home of hit shows including The Walking DeadMad Men, and Breaking Bad, noticed these changes early and took the opportunity build up a strong machine-learning foundation that network marketers could use to better understand and meet the needs of the audience.

IBM Big Data and Analytics Hub recently sat down with Vitaly Tsivin, executive vice president of business intelligence for AMC Networks, to gain a better understanding of how AMC set itself up for success with machine learning.

Establishing a foundation for machine learning

In our discussion, Tsivin was clear that machine learning isn’t something that can be implemented overnight or built in isolation. He noted that there are several pieces of the puzzle, including a diverse array of data, the right infrastructure to manage that data, data models and personnel.

“It all starts with creating your own data,” Tsivin says, making sure to follow that up by expressing the importance of sourcing data from third parties. This third-party information often takes the form of Nielsen data, providing viewership numbers and other trends. This broad foundation of internal and external data is the first building block in building a strategy for machine learning.

But you need more than just raw data. According to Tsivin, “You need to have the best tools available and the best platforms available in the market.”

These include both data appliances and cloud solutions which each bring something unique to the table. Cost effectiveness and speedy processing of data are just two of the benefits of appliances Tsivin mentioned.

Cloud, on the other hand, was characterized as a “hands-free”, elastic solution that lets users expand according to their needs without needing to procure space or resources in one’s own data center. When brought together as part of a hybrid implementation, organizations can establish a robust data management infrastructure flexible enough to handle the demanding needs of machine learning.

Lastly, one must consider hiring and business intelligence. During the discussion, Tsivin mentioned “forecasting and predictive models” as something that extends into machine learning and artificial intelligence. With that, an organization needs more people well versed in data science and analytics. These roles will include people with engineering or statistics backgrounds who “understand the entire spectrum of optimization and forecasting algorithms” to “formulate the task of machine learning.”

The solutions AMC uses

Tsivin said that AMC was “well positioned to bring machine learning into the picture” thanks to the IBM data warehouse appliance, IBM Integrated Analytics System, “one of the best in the market.” Similarly, he noted that IBM Db2 Warehouse on Cloud, was “one of the best ones there is” due to the rich functionality of the established Db2 offering, now combined with new cloud technology “such as in-memory execution and elasticity.” Specifically, when compared to Amazon Redshift, Tsivin indicated that “there are many aspects of that in which IBM is ahead.”

What really brings out the best of both solutions is the way in which they work together in a hybrid framework. Having the same technology and code structure underlying both introduces important benefits. “Your optimization algorithms run the same way and optimize the same way”, Tsivin observed. In addition, he also remarked upon the ability to move a lot of data between the two solutions, as well as how training employees becomes more cost effective due to a lower learning curve.

What AMC saw upon implementing machine learning

With a solid data management foundation, AMC was able to implement machine learning and has already begun to see the benefits, Tsivin said.

Foremost, it has helped it to “be in step with the way audience viewing is changing”, helping the organization to react quickly and optimize promotional campaigns. It’s also saving on “very expensive and very hard-to-find labor” thanks to machine learning. Instead of having someone look at and rebuild data models monthly or quarterly, AMC relies on machine learning to identify patterns and do its own rework.

How will your business use machine learning?

Of course, these benefits aren’t relegated to the realm of television and entertainment. Organizations in every industry have the potential to bolster success with machine learning. For example, in banking, we have customers using our IBM Integrated Analytics System and Db2 Warehouse on Cloud or Db2 Warehouse that are finding incredible value in more accurately identifying patterns of fraud, then acting on them in the moment.

Supply chain customers are also seeing value in quickly identifying patterns of the best routes to take while those in healthcare are working to improve treatment by recognizing patterns across numerous vital signs and test results. No matter which industry you’re in, taking steps toward machine learning now has the potential to pay significant dividends later.

Learn more about AMC’s story and the solutions it used by taking a look at the full case study. For a more in-depth discussion about the IBM Integrated Analytics Solution that AMC uses, read IDC’s latest report.