Predictive Analytics – Targeting Audiences & Delivering a Contextualized Experience

CMET_BlogImage3In simpler times, media companies operated within the competitive but predictable confines of controlled distribution models. Publishers, for example, thrived by building their subscriber base and by offering advertisers access to attractive demographic targets. Broadcasters and cable companies developed programming that yielded high ratings, and advertising revenue was relatively steady.

Today, of course, things are less straightforward. Consumers now access content and entertainment via a variety of channels and forums – mobile apps, broadcast and satellite programing, social media, multi-channel networks via YouTube and a wide variety of other consumption ecosystems.  Rather than offering audiences a long-form article and/or programing, modern media must cater to a “Gen-Z Generation” with a 3- to 5-minute attention span.

In this environment, the task of media organizations fundamentally shifts from pushing content out to audiences to pulling audiences in to provide access to specifically defined and relevant content.  This, in turn, creates a critical need for targeting audiences and delivering a contextualized experience. For example, insight into how, when and where consumers access content is becoming key. And the “where” part of the equation involves not just geographic location but context of how the consumer engages with the content – all of which can impact consumer behavior and interests.

Gaining that audience understanding, meanwhile, requires gleaning nuggets of insight from mountains of unstructured and/or inaccessible data residing on a myriad of disparate, disconnected and discrete internal and external sources.  

And for marketers, therein lies the challenge.

Specifically, where do you begin to look for insights from customer and market data? How do you organize various data sources and draw meaningful connections between them? What outputs can be created that drive informed and measurable decisions?

To address such questions, marketers are increasingly teaming up with their technology teams and technology service partners to develop solutions and execute strategies aimed at organizing data infrastructure and enabling data migration. Additionally, by utilizing Artificial Intelligence (AI)-fueled modeling and leveraging predictive analytics for audience insight, media companies can deliver a relevant contextualized content experience.

The opportunities are significant – machine learning systems can run hundreds of “what if” models in real time, and deploy pattern recognition capabilities to identify links and correlations between seemingly random data sets. Such analyses shed light on audience preferences and how consumers respond to different types of content, delivery channels and external factors. For example, audience interest can be impacted by variables such as the weather, the time of day and events ranging from the Super Bowl to terrorist attacks. Machine learning systems can gauge the impact of these variables, as well as assess the relative significance of a given variable’s level of influence.

The obstacles to achieving this vision, meanwhile, are daunting. Organizational issues perpetuate the existence of operational silos that prevent data sharing and integration. Inadequate infrastructures further hamper data transparency, and hobble the effectiveness of machine learning tools. Up to 80% of all Big Data is unstructured and unusable. As a result, media marketing teams today struggle to get their arms around available structured data, as well as access and utilize potentially valuable unstructured data. Existing analytical tools lack predictive and machine learning capabilities, are typically based on limited slices of data and are unable to provide the critically needed real-time 360 degree views of an audience.

Problems with disparate platforms are another issue: according to, almost 5000 solution providers are today creating silos of data, often with no single source of truth.

So how can you overcome these obstacles and realize the potential to use predictive models to get smarter with your audiences? In the coming weeks and months, we’ll outline key elements of an operational strategy designed to support the business imperative of achieving meaningful audience insight. We’ll examine how to address organizational and operational constraints, as well as define critical success factors and outcomes related to deploying machine learning. Themes will include benefits of integrating disparate platforms and data sources, leveraging AI and machine learning and getting deeper insight into campaign performance.

Send me your ideas for topics and examples you would like to cover.