Produced in association with Google
While digitization has strained the marketing department’s traditional methods and measures in recent years, the function is regaining its energy by plugging into a rich and abundant power source: data.
Existing enterprise data, as well as information gathered from engaging with consumers, has become a valuable business asset. With the advent of machine learning—a type of algorithm that identifies patterns in data and improves with experience—companies can use data to predict and “learn” to identify consumers who appear likely to become high-value customers.
The business restructuring taking place today mirrors the transformation that’s underway throughout entire industries, from media to manufacturing, as companies rewire to compete in a digitally drenched environment. “For a long time, none of us would have thought that selling food and beverage over e-commerce would have been such a large business,” says Shyam Venugopal, vice president for global media and consumer data strategy at PepsiCo. “We are operating in an environment that is constantly changing.”
Giving the growing complexity of customers’ buying journeys, which now span multiple devices, platforms, and touchpoints, the task of keeping up can seem overwhelming. Marketers have had to radically rethink their approaches to reaching potential customers in an era with infinite options for creating and distributing content. Their goal is even more ambitious: to accurately anticipate their potential customers’ next moves, capturing and even shaping what they want and need.
The profusion of data that companies can collect represents a wish-fulfillment opportunity for marketing executives at big companies yearning to return to building one-to-one relationships with customers. By investing in and harnessing machine learning techniques to comb the data for insights, identifying patterns and creating predictive models, marketers can map the customer journey. They can gain an understanding of how different segments make buying decisions and present them with more personalized messages.
Seizing the data
A recent survey conducted by MIT Technology Review Insights, in association with Google, confirmed the centrality of data in building a hyper-personalized marketing strategy. The survey canvassed 1,419 marketing executives from companies with more than $100 million in annual revenue, from various industries, including financial services, retail, technology, and education.
After careful consideration of business metrics, respondents could be categorized as “leaders” or “laggards”: Leaders represent companies that achieved a greater-than 15 percent increase in revenue over two years or a greater-than 15 percentage-point increase in market share over the same time period. In contrast, laggards were companies that saw shrinking revenue or a loss of market share in the same time period. Among respondents, two-thirds of leaders say how companies apply their data will play a key role in their ability to thrive.
In addition to the technical bridges that must be crossed to extract actionable insights from data, functional barriers also need to be lowered. The survey found that leaders are 60 percent more likely than laggards to believe the marketing team should own a data-driven customer strategy that supports all organizational stakeholders. In the past, marketing may have overseen demand generation, but it passed those leads over to sales. Now marketing needs to take ownership of the entire experience, spreading insights and analysis throughout the business.
Designing the path of no resistance
With its ability to understand and anticipate the most effective marketing approach for each current and potential customer, machine learning relieves marketers of the heavy lifting associated with downloading and manipulating massive quantities of data.
Consider, for example, a marketing function that has amassed 2.5 million e-mail addresses and wants to contact people who are most likely to be receptive to a new product offering. Manually segmenting that volume of customers is time-consuming, but spraying them with spam will generate “unsubscribes,” and re-acquisition is prohibitively costly. Machine learning can efficiently extract e-mail addresses whose owners have, for example, a 25 percent chance of opening the e-mail and a 1 percent chance of unsubscribing. Guided by machine learning logic, marketers can identify and mix the optimal combination of elements that will likely lead to a successful business objective.
Tracing the impact throughout the business
Consistently executing on that level, however, requires more than layering in new technology. Maximizing customer-centric processes means unifying them, tying disparate systems together and putting them at the center of the business. For companies typically organized by geography or product line, the upheaval represents a DNA-level cultural change—a new corporate architecture that supports customer-experience management.
Where might such an effort begin? “In each part of the organization, the definition of what a customer is may be different,” observes Allison Hartsoe, founder and CEO of Ambition Data, a data analytics consulting firm. “The organization needs to create a unified view of the customer—a larger, broader definition that everybody benefits from.”
The commitment to becoming customer-centric and data-focused invariably requires organizational restructuring. “It becomes necessary because the customer data is spread across so many departments; so to be comprehensive you need the organization to align,” says Hartsoe. “It’s about changing the way the entire company thinks.” At some companies, achieving that goal means adding a chief analytics officer to the C-suite.
In any case, transforming into a strong metrics-driven organization requires discrete functions to work together much more closely. At one gaming company, says Hartsoe, product managers don’t make choices about which product features to add without getting input from the customer research group. Given the company’s decision to focus on high-value customers, “before any choice is made, the product managers want to wash it through the metrics of high-value customers, so they can make decisions in a customer-centric fashion,” says Hartsoe. “Using metrics changes the process of decision-making.” It also enables companies to pre-emptively assess the impact of any product iteration under consideration.
Equipped with machine learning technology, marketers will not only do their work differently—they will also do different work. In the survey, nearly three-quarters (73 percent) of marketing leaders investing in machine learning say they have shifted more than 10 percent of their time from manual activation to strategic insight generation. They can now more diligently manage their investment-allocation decisions, devoting their limited resources to the most promising potential customers. Machine learning can compare historical data about people who became buyers to newcomers who show similar pre-purchase behavior. The result: a predictive model based on a customer’s future intent.
As the survey found, leaders are 53 percent more likely than laggards to say machine learning processes data signals to help marketers better detect consumer intent. The ability to collect, organize, and analyze data from multiple sources enables marketers to pinpoint how close customers are to a buying decision, shaping messages that are relevant rather than intrusive. Modern consumers, whose expectations of what they can buy when—namely, anything and right now—prefer doing business with companies that understand and assist them.
An accumulation of such customer-positive experiences forms the building blocks of what might be the most durable competitive advantage: customer loyalty.
Winning over customers, capturing returns
Marketing departments can now sort customers into groups with better accuracy. By defining value-based customer segments, companies can boost their marketing return on investment by timing delivery of their messages when the data indicates that potential customers are on the verge of making a decision.
Machine learning can also measure how effectively marketing is investing its budget, continuously calculating the return. The survey found that leaders are twice as likely as laggards to agree that using machine learning in media campaigns has improved worst-performing ROIs by 10 percent or more. Marketers can measure the impact of using different marketing tactics while also quantifying the value of isolated elements within each of those activities. With that in mind, they can reallocate their budgets, shifting away from the least efficient channels.
By uniting disjointed data silos and applying machine learning to all of the organization’s data, marketing will gain, and keep, a clearer view of those customers than ever before and drive growth for the business. As marketers develop confidence in the added value that results from integrating machine learning into their function, a growing number will be converted into enthusiastic students of the technology—a priceless commodity in the predictive, digital age.
Learn more about machine learning’s impact on marketing.