Hearing What Isn’t Said

Thursday, June 15 2017 Aaron Harms EVP, Product and Technology Categories:

How natural language processing at Protagonist exposes the belief patterns lying below the surface


As the pollsters charged with predicting the 2016 election results can tell you, understanding belief patterns at scale is hard. While most businesses aren’t trying to wrangle a group as broadly distributed as the American voting public, they’re still often working to appeal to an audience of millions, and to do so they need to understand the Narratives at play.


Unfortunately for the enterprise world, consumers are rarely so kind as to lay out the exact belief systems that drive their decisions. Instead, they’ll make a specific statement or offer an opinion, with the Narrative behind it hiding between the lines. If a human sat down and read through every publically available comment made by a company’s target audience members, he or she would be able to piece together the puzzle of people’s beliefs-inflection, tone, whether the comment is making a subjective claim-all of these hint at the underlying belief. But no human has that kind of time, and those sentiments only gain meaning when taken as a whole.


At Protagonist, we’re focused on building new Natural language processing (NLP) capabilities to identify and make sense of the clues that reveal consumers’ underlying beliefs. NLP is everywhere, powering chatbots, Siri, and search and much more. And at the core of every NLP geek’s ambition is a desire to build algorithms that can grapple not only with the syntax of language but also the semantics-the underlying meaning. Narratives are powerful because they tap into that underlying meaning-and that’s also why they are so hard to pin down at scale.


Think about the kinds of language that friends use to make an impassioned arguments to each other about causes they believe in. One will make claims, back them up with evidence, and the other may see some of that evidence as factual and some of it as baseless. One might deploy analogy and metaphor-attempting to relate his or her argument to other arguments that the other may be more familiar with or already bought into. They’ll punctuate their points with emotion and tone-appealing to their sense of empathy in order to place emphasis on the crux of the belief. This is just a sampling of the types of Narrative patterns that Protagonist mines and analyzes.


The Protagonist Platform extracts these patterns at their root and exposes trends for our analysts to evaluate. Sometimes, we find Narratives that are unexpected or counterintuitive. Other times we find that new Narratives have caused a shift in the conventional wisdom about a certain market. These aren’t findings that one would get by looking at keywords, basic topics, or standard sentiment-they’re grounded in a data-driven understanding of belief and meaning. Whether workers are charged with understanding and activating a target audience, energizing and protecting their brand or shaping a conversation, tapping into Narratives with data and at scale will drive the campaign that resonates, an enduringly healthy brand and the confident messaging that wins audiences over.