If you don’t understand machine learning (ML), it’s time to give yourself a crash course—especially as it relates to marketing.
Just as SEO, programmatic buying, and other new technologies have revolutionized certain aspects of the marketing industry, ML figures to change the whole process, from how we handle simple marketing tasks to how we approach the larger task of telling brand stories.
First, though, it’s important to dispel common misunderstandings. ML isn’t the same as artificial intelligence. Instead of developing the cognitive capabilities to rival or even surpass human intellect, ML is more about optimizing certain problem-solving processes.
ML, in other words, is just a very advanced tool. But it’s a tool that could change everything because of how quickly it can work, and how much complexity it can handle. Data is critical to an effective ML strategy, which is where it dovetails nicely with today’s digital marketing strategy. Digital marketing’s best opportunities all flow from data, and marketers will quickly discover that their ability to leverage ML technology will only be as good as the data they have driving those insights.
While marketers may not realize it, basic versions of ML have been used for years. And marketers are gradually building more sophisticated digital strategies around the ML technology available today. The impact has been powerful, but it’s nothing compared to what’s on the horizon.
ML is not necessarily a new tool—it’s just something that has evolved over time, and gained considerable new strengths in recent years. Spell-check tools have long used basic ML principles to identify spelling and grammatical errors. The tools weren’t always perfect, but they used basic data sets to recognize potential errors.
More recent examples include online product recommendations, which use data to make algorithmic suggestions to consumers. Google’s search browser also routinely features examples of ML: Not only do search results themselves flex the power of machine-centric problem-solving to understand intent and speak to the searcher’s pain points, but Google routinely uses ML technology to make sense of search queries, such as when those queries are littered with spelling errors.
Of course, newer ML technology is even more robust and accessible to non-tech giants. Companies like Sift Science use ML to detect online fraud, while IBM’s Watson solution combines ML with natural language processing to provide assistance across a range of services. Access to data, combined with evolving ML tools, has opened the door to innovative new applications in marketing, and brands are taking advantage.
Brands are already using ML to drive insights and operations that go well beyond text. SailThru, for example, uses ML to make brand email marketing more efficient and productive. Machine learning analyzes consumer behaviors to determine when email delivery is most likely to draw engagement and conversions. By timing emails according to those insights, SailThru’s clients have seen double-digit increases in email-generated revenue, according to Harvard Business Review.
And as behavioral and contextual data becomes more available to brands, ML technology is making it possible to make use of that information. For as valuable as data can be to marketers, it’s useless if marketers are unable to put that data into context and drive insights from the information. ML picks up where those marketers and their analytics tools fall short, processing data at an enormous volume to leverage that information for valuable insights.
One of the most infamous examples of this came a few years ago, when Target’s marketing team learned of a teenager’s pregnancy before the girl’s parents. That realization came by examining her shopping history and giving it context based on past purchases from other consumers. Based on her shopping behavior, Target reached the conclusion she was likely pregnant, and the company shifted its marketing to specifically address her new pregnancy needs.
That’s a prime example of how ML works to drive insights previously out-of-reach. Target’s massive consumer base, and its even larger set of shopping histories, presents analytics needs far beyond what human staff can manage. But ML is able to handle the workload, and it allowed the retailer to drive incredible consumer insights.
ML is already being utilized in marketing campaigns around the world, but the full force of its influence is yet to arrive. As Adobe points out, ML will soon move from its service at the individual consumer level to a larger role that considers large groups at once, and external factors as well as internal data.
Adobe suggests that weather and competitive analysis, for example, will be on the table for ML to consider alongside a wide range of other factors. And predictive analytics publications like insideBIGDATA believe that ML will revolutionize storytelling strategies online, with data-driven storytelling representing the next wave of analytics applications. That’s already available to marketers, albeit in a restricted form: Data can identify the subject matter, but the story still has to be built by marketers.
Over time, the balance of that workload may shift. ML could eventually dictate not just the subject matter of content, but also the right medium, the tone, and other critical facets of how brand stories are told. It could drive immersive experiences based on interactive content: Basic ML components could be built into interactive content, creating a personalized experience that leverages a combination of in-house data and user feedback to customize real-time content on a one-to-one level. A short quiz to help consumers define their fashion preferences, for example, could use answers from past questions to refine recommendations much more quickly and accurately, and to present more relevant questions than ones presented through a standardized quiz.
Eventually, ML could also be used as a tool to drive non-linear brand storytelling, or to evaluate and maintain narrative arcs and brand voice. It isn’t hard to imagine a brand building a complex narrative over a large span of time and across a variety of mediums: In this scenario, ML could ensure that connecting pieces of content match one another in tone and fit into the rest of the content. ML can essentially test the countless paths consumers may take through content, crossing multiple channels in the process and make sure that each step is synced with the one before and after it. That’s a process that would take human users far too much time to do on their own, and it would be rife with human error if they did.
The long goal is using this technology to create better content and tell better stories—human content still, but content that is more effective and more relative to the company’s cause.
So no, the machines are not plotting their takeover. As far as marketers should be concerned, ML only hopes to make your job easier.
To read more about how technology will change brand storytelling, subscribe to Content Standard updates.