Analytics-driven targeting, personalization and engagement in the insight economy

Big Data Evangelist, IBM

In the insight economy, predictive audience analysis is an essential component of customer targeting, personalization and engagement. As the media and entertainment industry grapples with the disruptive impact that online, digital channels have on its core delivery model, audiences have become increasingly amorphous. Television audience measurement, in particular, is still trying to adapt to this new industry order: digital, online, on demand, interactive, multimedia, multiscreen, streaming, mobile, social and so on. With a revenue model that remains indelibly advertiser supported, the TV industry continues to introduce new tools that capture the diverse facets of viewership—or, broadly, engagement in the interactive sense—in this new world.

In spite of all the high-powered tools to get our finger on the pulse of our audiences, pulling this measure off consistently remains fiendishly difficult. In that regard, pity the poor Hollywood executive—ignore the money, power, glamour and Bel Air mansion for the sake of argument. Screenwriter William Goldman had the classic observation about the hubristic hotshots who bankroll major motion pictures. Though everybody in Tinseltown likes to pretend that they have their finger on the pulse of the moviegoing public, in fact, said Goldman, “nobody knows anything.” Every time a mega-production flops at the box office, every time a surefire sequel lays an egg and every time a consistently reliable star cannot put butts in the seats, the same sad refrain occurs. Watching millions of dollars go straight down the tubes, the studios all say “never again,” until, that is, they bet big bucks on the mega-turkey, and the one after that, and so on and so forth.

Challenges in predicting demand

The personalities of an audience—some refer to it as zeitgeist—are what change like quicksilver, making predictive analysis sensitive to myriad factors that can’t be modeled in any straightforward fashion. Of course, the entertainment industry isn’t the only sector that has trouble predicting demand. Even with predictive analytics and social-listening engines at their disposal, most industries need to handle the intrinsic fickleness of taste. People love what they love until, for whatever reason, they don’t love it all that much anymore. Randomness, restlessness, curiosity, boredom, satiation and other taste-shaping factors may gang up to suddenly render yesterday’s mania hopelessly passé. Even if we’re paying close attention to these rapid shifts in popular demand, the underlying causes may not be obvious in any way.

Nevertheless, getting an acceptable handle on audience personality profiling is possible with the right data-scientific tools, especially cognitive computing and streaming analytics. To support customers in diverse industries, IBM announced last year its new consulting practice to address experience management within a comprehensive audience-targeting and engagement strategy. At that time, IBM introduced three key experience-focused algorithms that benefit greatly from high-quality social, mobile and other new audience data sources:

  • detection: This algorithm analyzes unstructured social media data to detect important events in customers’ lives. For example, this technology can assess specific life events such as a marriage and then make correlations to a range of financial decisions.
  • Behavioral pricing: By combining behavioral models on consumer response to pricing, such as surprise and thrill of a deal, with historical transaction data, this algorithm helps retailers design personalized pricing strategies that help consumers make purchasing decisions and improve their experience.
  • Psycholinguistic analytics: By combining the psychology of language with social media data, this algorithm helps in understanding inherent personality traits of individuals and identifying their preferences. This technology goes beyond generalizations and recognizes the individuals to identify how they prefer to receive and consume information and offers.

Predictive personality profiling

Personality profiling took a big predictive step earlier this year in the form of a partnership involving IBM Watson and development studio Chaotic Moon based in Austin, Texas, which was reported in a recent Data Informed article. IBM and Chaotic Moon are teaming to gain insights on individual audience members’ personalities in real time. The team is using Watson Personality Insights on the Bluemix cloud platform to derive personality and social traits through automated linguistic analysis of Twitter-based customer communications.

This technology can be used for a real-time immersion in audience analytics, tailored messaging and response metrics. It leverages Watson’s concept expansion, message resonance and relationship extraction features to personalize and contextualize audience engagement strategies in real time. The tool can infer cognitive traits, social tendencies, basic values, fundamental needs, situation-specific reactions and what the referenced article refers to as “emotional resonance” of specific messages with various targeted recipients.

This sort of tool can be used for one-to-one personalization, based on classification into fine-grained personality types aligned with various propensities and preferences for engaging with media, brands and communities. It does so by incorporating customer data from social-listening tools and other sources, enabling you to build and test myriad personality profiles that reflect people’s various behaviors in mobile, social, web and other channels.

Preparation for the insight economy

Learn how to leverage data science, cognitive computing and streaming audience analytics to drive the insight economy. Register today for IBM Insight 2015 in Las Vegas, Nevada, October 26–29, 2015.