Recently, author Shekhar Deo wrote an opinions piece about the idea of using machine learning to connect marketers with consumers.
The post gave an excellent primer into machine learning—how it "works with algorithms to analyze consumer behavior in real time to perform predictive segmentation and adapt the communication experience"—and how machine learning can be applied in the context of serving personalized ads.
Today, I'm extending the discussion to another realm: content marketing.
The Content Engagement Problem
B2B marketers are using content to meet buyer needs, and it is proving effective in engaging, nurturing, and converting prospects. Yet content engagement appeared at the top of the list of challenges facing content marketers with 54% of B2B marketers saying that they are challenged or very challenged in this area, and 50% of B2C marketers saying the same thing, according to MarketingProfs and the Content Marketing Institute.
The reasons are not surprising.
Although we're keen to shape the traditional buyer journey as a smooth, uncomplicated funnel, in the digital realm, the journey is much more complex:
Moreover, we are now in the "Big Content era," in which myriad pieces of content are being produced by every organization in the form of daily tweets, Facebook updates, blog posts, product offers, customer reviews, testimonials, product and service specifications, FAQs, and more. All that content need to be categorized and structured before being served across multiple platforms.
The B2B content marketer—and his or her colleagues in demand generation—must now become part-data scientist, part-content creator, part-knowledge manager, and part-psychologist to both understand each buyer and make the most effective message to serve to the right person at the right time.
How Natural Language Processing Helps in Content Creation
Natural Language Processing is a type of machine learning that reads and measures text-based content.
It analyzes text and identifies the topics and themes contained within it in the same way that you are reading this article and can understand the difference between nouns, verbs, and adjectives on the page.
Natural Language Processing doesn't just read and understand content; it also labels it with metadata (data about data). This metadata usually describes key elements, such as the author, topics, and content length, but it can include other information, such as the media type, purposes, sentiment, device it is intended for, and much more.
Applying NLP to Match Content Against Buyer Interests
The content we consume as buyers is hugely indicative of our current needs and evolving interests. Think about it: the fact that you clicked on this article already tells us that you're at least slightly interested in the topic of machine learning.
One article alone isn't enough to build an accurate picture of you, the reader, but if we were to track your reading arc around MarketingProfs, very soon we could build up an increasingly accurate picture of which topics are interesting to you and use that to identify your current concerns and needs (say, "machine learning," "B2B marketing," "marketing automation").
This kind of insight is hugely useful whether for a sales rep ("I know these people have all regularly read about 'machine learning,' which means I should call them about our latest paid report on that subject") or for email segmentation ("Let's send emails about our demand generation workshop to people who are interested in 'marketing automation'").
Why Machine Learning Works
At the risk of the above seeming hugely theoretical, let's look at how popular brand Kraft has made it a reality.
Last year at Content Marketing World, I had the pleasure of hearing Julie Fleischer (Kraft's director of Data, Content and Media) talk about the wins the company experienced as a result of using this approach to connect content to consumers.
The team at Kraft tagged and tracked more than 22,000 different attributes of its audience member based on their behavior and engagement with Web content.
The result: Kraft now generates the equivalent of 1.1 billion ad impressions a year and, according to Fleischer, garnered a four-times-better return on investment through content marketing than through targeted advertising—all because the company used machine learning to create a "content engine" that matched content to the right consumer based on his or her evolving interests exposed by content consumption.
* * *
The onus is now on marketing professionals to get wise about applying machine learning to their content marketing practice to make their content intelligent.