Fortune-Telling Goes Hi-Tech: The Rise of Predictive Analytics
Fortune-telling is a familiar scene in movies. The protagonist sees a neon sign with the word “psychic.” They stumble into a small shop on a crooked side street, motivated by idle curiosity, romantic uncertainty, or existential desperation. They push through an iridescent beaded curtain dangling in the doorway. An older woman, wearing a headscarf with coins, gazes into a crystal ball. Magical powers are summoned. A fortune is told.
Sometimes, that fortune turns out to be accurate. In such cases, skeptics might argue that the process utilized was anything but magical. It was a prediction based on a known set of facts. Customers reveal details about themselves through micro-expressions, vocal intonations, fashion choices, and of course, the information offered up in response to cryptic questions. The fortune-teller analyzes that data and, arguably, factors it into the prediction. She may not even be aware that she's conducting data analysis because the conclusion might feel intuitive or otherworldly to her. According to Malcolm Gladwell's bestseller “Blink,” the adaptive unconscious is a powerful tool, offering up quick analytical output in the form of intuition and gut instinct.
With predictive analytics, fortune-telling goes high tech. Mystical clairvoyance becomes technologically plausible. Algorithms crunch numbers and compute probabilities, determining consumer behavior and buying patterns. Businesses and marketers can then capitalize on this predictive capability.
Last year, Harvard Business Review surveyed 490 executives and managers to find out how AI has been affecting businesses. 82% of respondents said that predictive analytics produced the greatest impact on their organizations to date.
In an interview for this article, Stewart A. Skomra, a new product and new market development expert, commented, “Predictive Analytics has been around as long as humanity has kept a record of things and chose to plan-its-work and work-its-plan. Historically, Predictive Analytics – the application of algorithms to determine an expected outcome at a future time – were not easily automated.” Skomra observed that this has now changed, thanks to exponential advancement in semiconductor technology.
Predictive analytics can influence lead nurturing and illuminate target markets. It can inform personalized product recommendations, promotional offers, customer retention tactics, and credit eligibility. Its usefulness is dependent upon robust customer data and trained data analysts. A Mu Sigma white paper draws a distinction between predictive analytics and descriptive analytics, inquisitive analytics, and prescriptive analytics. According to that white paper, “all four kinds of analytics have to be done in the right mix.”
Some may resist the notion that their own behavior can be predicted by a computer, due to concepts of free will, individuality, and so forth. There will certainly be instances in which algorithms are wrong and it's important for executives to be able to study those faulty conclusions by using AI-powered tools with built-in transparency. However, there are also clear patterns of behavior that emerge when big data is properly analyzed. Predictive analytics may never arrive upon a point with absolute certainty, but it provides an educated guess and can consider more variables than a human. For the purposes of commerce and marketing, that equates to an actionable insight. It's enough to ensure that a marketing spend isn't a catastrophic failure.
Bernardo F. Nunes is a data scientist at Growth Tribe Academy. His work covers several disciplines, including natural language processing, deep learning for image recognition, supervised learning, and unsupervised learning for segmentation and to uncover psychographics. Last October, Growth Tribe Academy released an educational video titled “Predictive Analytics in Marketing.” I asked Bernardo some questions about predictive analytics. He mentioned the pirate funnel as one area of relevance.
“Basically, with supervised learning, we can train models that predict the steps of the funnel,” he said. He mentioned that predictive analytics can be used for lead scoring, churn prediction, and lead nurturing based on customer lifetime value.
“What I call a byproduct of the predictive analysis is the discovery of relationships between some of the features and the outcome. For example, more engaged customers are less likely to churn and generate a higher CLTV. Now, I can A/B test different nudges that try to improve engagement on the website,” he explained. He mentioned that blog posts and email marketing are some cost-effective ways to impact the desired step of the funnel. He continued, “This is the integration between machine learning and rapid experimentation of a business. Probably, we will find higher uplifts with this integration.”
Bernardo added, “Finally, I would not forget unsupervised learning, which is used in business for customer segmentation but is not predictive analytics. You want to create content and marketing strategies that fit your segments. You discover them by using K-means and/or hierarchical clustering algorithms. They basically measure how similar your customers are based on their characteristics (socio-demo, behavioral) and group them accordingly in K groups. Your marketing team will then develop content that fits young vs. older, high CLTV vs. low CLTV, engaged vs. non-engaged customers, et cetera.”
Predictive analytics is just one part of a new technological trend towards efficiency. According to a global analysis of venture funding produced by KPMG Enterprise, this innovation is well-fueled. “In 2017, VC investment in artificial intelligence almost doubled, attracting $12 billion of investment globally, compared to $6 billion in 2016,” the report states. These new technologies could fundamentally change our very impressions of capitalism. Companies are becoming more and more efficient at creating products that people want and connecting consumers with those products.
Although Kickstarter may seem technically primitive when compared with some of the rapid, transformational developments in AI, it offers a means for entrepreneurs to test whether there is public interest in their product proposals. Many people think of crowdfunding platforms as a method for circumventing bank loans and venture capitalists, but these platforms also assess wants and needs in an unprecedented way, condensing timelines and reducing risk. By testing and reacting, entrepreneurs are spared from the extreme hardships of bad investment. Similarly, brick-and-mortar retailers have been gathering consumer insights by tracking shoppers' movements with floor sensors and studying geolocation data from smartphones. They analyze which products are considered but never purchased and which aisles are rarely visited. With all of this data, they can stay in business and improve profits by designing better store layouts and reevaluating product displays.
Taken as a whole, there is potential here for an altered relationship between companies and consumers. Customers will gravitate towards businesses if they have efficient systems for assessing wants and needs and then optimizing pricing, presentation, and delivery. Marketing will also become increasingly data-driven, whereas in the past it was more of an exercise in speculative creativity.
In the TV series “Mad Men,” Don Draper and Peggy Olson spitball ad campaign ideas just to see what sticks. The objective is to convince, not to problem-solve per se. In the era portrayed by the show, there was a necessary element of manipulative persuasion, a twinge of trickery to the process of buying and selling. Creative decisions were highly subjective. Product features were exaggerated or distorted. Integrity sometimes went by the wayside. Just look at an old ad for cigarettes.
Yes, many of these marketing aspects still exist today and will endure for some time. But it is possible that this new wave of technological innovation will move commerce into the realm of efficiency and away from manipulation. Digital marketing will tell you about products that are actually relevant to you, satisfying your curiosity and addressing your exact needs before you have even articulated them. The modern generation of Don Draper's and Peggy Olson's won't go away just yet. The best way to create an ad that resonates with a human is, still, to have a talented human create that ad. However, the field of marketing is clearly changing. Data scientists in Silicon Valley are now as essential as the creatives on Madison Avenue. This new era rests upon emerging fields such as predictive analytics.