Data Mining's Future in the Media Industry
Do you remember the pictures that at first glance look like a jumble of colored dots? If you stare at one in just the right way, a three-dimensional picture jumps out from the background. Think of those dots as the bits of information about your customers contained in your company's databases. If you look at the dots of information in the right way, they reveal patterns that yield insight into consumer behavior.
Traditionally, organizations use data tactically to manage operations. For a competitive edge, strong organizations use data strategically to expand the business, improve profitability, reduce costs and market more effectively. Data mining creates information assets that an organization can leverage to achieve these strategic objectives. Some industries, including banking and finance, have stared long and hard at their customer data dots.
Two challenges. Media companies face two challenges that are having a fundamental effect on their basic business model: audience fragmentation, and declining efficiency and elimination of telemarketing.
In the past 20 years, we have had an explosion of content delivery channels. This channel proliferation creates a unique problem for media companies: How do we continue to aggregate the number of consumer "eyeballs" when consumers have so many choices for content?
With few exceptions, this proliferation has resulted in most mediums losing eyeballs. For example, newspapers have lost more than 10 percent penetration of weekday readers since 1980 and more than 20 percent since 1970. Worse, as the younger demographic ages, it is not assuming the media habits of the older demographic. According to a Newspaper Association of America study, only 26 percent of 18- to 34-year-olds read a daily newspaper. This is less than half the use level of 45- to 54-year-olds. And 18- to 24-year-olds are nearly as likely to use the Internet as a news source than read a weekday newspaper.
Meanwhile, media companies traditionally have relied on outbound telemarketing for subscriber acquisition and product upsell. But consumers today routinely screen calls using answering machines, caller ID and other technologies. Plus, marketable telephone numbers are being called at increasingly shortened intervals as a result of automated, intelligent, prefix dialing.
The result is that do-not-call lists continue to grow. Given these privacy-related trends and concerns, many experts predict that cold calling will disappear through regulatory constraint.
Data mining can help. Data mining can be used to overcome audience fragmentation and the decline of telemarketing.
Of course, companies always have tried to quantify customers' wants and needs -- and largely failed. Department store pioneer John Wanamaker famously complained 130 years ago that half of his advertising was wasted; he just didn't know which half.
Today, companies harness the power of data not only to answer Wanamaker's question but to go far beyond that. They mine the mountains of data to prompt customers to buy more, stick around and perhaps even pay extra for tailored products and services.
We define data mining as "the data-driven discovery and modeling of hidden patterns in large volumes of data." Data mining differs from retrospective technologies because it produces models that capture and represent hidden patterns in the data. With data mining, users can find patterns and build models automatically, without knowing exactly what they are looking for.
These models can be both descriptive and prospective. They address why things happened and what likely will happen next. A user can pose what-if questions to a data mining model that cannot be queried directly from the database or warehouse.
Examples include: "What is the expected lifetime value of every subscriber account?" and "Will this customer cancel the subscription if we raise our prices?" For media companies, this type of data mining allows for the identity of the optimal targeting mix as shown here:
Optimal Targeting Mix = [likelihood to respond, likelihood to stay, likelihood to upgrade, advertiser appeal]
Predictive modeling. Across large enterprises, predictive intelligence technology is used for a broad set of applications. Retailers forecast demand down to the store and item level. Drug companies use it to develop drugs, then figure out what marketing programs will cause doctors to write more prescriptions.
Do media companies use it as effectively? Most of the traditional business processes that companies use have been unable to draw correlations to external factors. With this newer predictive technology, a retailer that traditionally made forecasts based on last year's sales, for example, now can factor in external variables such as the opening of a competitor's store a few miles away. Or, a newspaper looking to predict single-copy sales might run a model incorporating weather, editorial content and even traffic patterns for the day along with the current circulation trend.
Many companies already reap big benefits by mining customer data. In a recent survey by Forrester Research, 34 percent of large companies expect more effective use of customer information to help them cut costs this year. For example, more newspapers use telemarketing and direct mail acquisition models to optimize their marketing spend. These models historically provide response lift compared with random marketing or the purchase of vertical lists.