Democratizing Machine Learning Algorithms for Integrated Data-Sharing
Media companies have come a long way in how they make critical business decisions, in the midst of a technology revolution unprecedented in its ability to challenge and change the industry. Thanks largely to stunning technological advancements in the last decade, surveys, focus groups, rating charts and rankers have been supplemented or supplanted by mountains of granular data, along with highly sophisticated collection and analysis techniques, that are transforming the way media companies produce, acquire, package and distribute content. The end result, arguably, is a better consumer experience, as well as increased advertiser value. This technological tsunami can be overwhelming, however, and knowing how to capture and utilize its power is a key to succeed in this turbulent industry.
ION Media Networks, whose flagship channel, ION Television, catapulted to a top 10 U.S. cable network in less than a decade, has taken a relatively straight forward approach with its technology strategy, and then deployed it meticulously throughout the company. First, we work to understand the current and future data needs of our key functions and then build an infrastructure to efficiently and effectively collect, analyze and internally distribute data from a multitude of sources. We promote a culture of integrated datasharing and training to arm decisionmakers with the best tools to help them best serve our viewers and advertisers. We promote a culture of integrated data-sharing and training to arm decision-makers with the best tools to help them best serve our viewers and advertisers.
We promote a culture of integrated data-sharing and training to arm decision-makers with the best tools to help them best serve our viewers and advertisers
As more companies operationalize research and analytics with the likes of addressable advertising, now at the scale that television demands, we leverage our data to prioritize our viewers, which, in turn prioritizes our advertisers. We also create competitive collaboration, which enables our employees to operate in a more intelligent and effective manner.
With a mix of primary and syndicated data sources and the help of methods from simple descriptive analytics to machine learning algorithms, we create a “data fingerprint” of an audience and the content that resonates with it. Our data strategy allows us to derive a start-to-finish view of consumers, from impression through purchase, in order to identify new and more specific consumer and content segments via patterns and features in the data. So, even as our data scientists bury themselves in building mathematical models, they also are actively accounting for the preferences of all our viewer segments within those models and eventually informing our decisions about inventory management and content acquisition. By focusing on providing our viewers with the programming they find most compelling and optimally stacking programs on our schedule (the television counterpart of digital recommendation engines) we are also creating significant value for our advertisers.
Organizationally, we do not separate research, traditionally tasked with analyzing syndicated data sources, from analytics, which applies data mining and optimization techniques to the data. By joining the information and methodologies from both disciplines, we are able to create a holistic view of consumers and content, which empowers decision-makers throughout the company. Ultimately, intensive collection and analysis of viewer data allows us to continue to attract and maintain viewers while taking calculated risks, creating new business and outperforming the market.