Zero One: What Are Your Customers Really Thinking?
Silicon Valley’s dirty little secret: Tech workers aren’t really concerned about diversity and gender equality, rather they care more about the bottom line and feel they’re being force-fed diversity.
At least that’s the “real story” uncovered by tech vendor Protagonist’s narrative analytics software, which looks at millions of pieces of content, from published stories and blogs to social media comments, and boils down conversations to come up with thematic narratives.
With a machine learning system doing the analysis, Protagonist claims, these narratives are free from political correctness or a single person’s agenda. In so doing, Protagonist hopes to get a sense of what people are actually thinking.
“It’s often not the case that there’s some big secret out there, but you get an understanding of specific language, what’s actually triggering people,” says Protagonist CEO Doug Randall. “We can tell companies, this is the belief system you can market to.”
Narratives that provide insight – even imperfect insight – into people’s feelings about products and services can be valuable to companies. This is evident by Protagonist’s impressive client roster: Starbucks, HP, Wells Fargo, Pfizer, Salesforce, Department of Homeland Security, and Microsoft.
Microsoft, for instance, wanted to market to small and medium-sized businesses, or SMBs, more effectively.
Protagonist’s narrative analytics software found that SMB executives view Microsoft as offering basic plumbing services to run their businesses. Not particularly insightful nor helpful. But the narrative analytics software also uncovered a variety of SMB attitudes that could help Microsoft foster better relationships.
One of the narratives showed a segment of SMB executives with big aspirations to grow their businesses. This led Microsoft to create marketing campaigns and webinars appealing to entrepreneurial dreams, such as ways executives can become heroes to their customers.
In stark contrast, another narrative showed a segment of SMB executives with lots of fears and concerns about their businesses. This led Microsoft to create marketing campaigns and webinars around solving real problems, thus alleviating some of those fears and concerns.
“By Microsoft’s account, that led to a four-times growth in their pipeline due to their marketing and sales efforts,” Randall says.
It should be noted that Protagonist bills its narrative analytics software as machine learning, but this might fall short of machine learning definitions, which would debunk some of the accuracy of the narratives.
True machine learning delivers relatively accurate results because it happens without human intervention and isn’t dependent on potentially flawed hard-wired programming. A machine learning system also requires reinforced learning – as in, what was the outcome? – in order to evolve and become better at achieving its objective function.
But narratives by their nature are subjective. It’s impossible to say a narrative is accurate or inaccurate, which means there’s no reinforced learning. That’s why Protagonist must insert humans into the equation; people look at the software-produced narratives and essentially give them a thumb’s up or down.
From a business standpoint, though, the fact that Protagonist’s narrative analytics software doesn’t produce the kind of definitive insight possible through machine learning might not matter.
Companies are starting to understand that narratives in a socially noisy, easily shareable world have a major impact on sales and profits, and they need to get out in front quickly even with imperfect information.
Falling behind a narrative can have dire consequences.
“Uber would say, gee, we have a little bit of a narrative problem,” Randall says, “with our customers, our employees, with everyone.”
Tom Kaneshige writes the Zero One blog covering digital transformation, AI, marketing tech and the Internet of Things for line-of-business executives. He is based in Silicon Valley. You can reach him at firstname.lastname@example.org.