Have you ever given much thought to how a frog boils to death? Conventional wisdom has it that if you put a frog into a pot of boiling (or very cold) water, it’ll jump out having noticed the instant change in its environment. If the frog is put into a pot of room-temperature water that is heated up slowly, it’ll happily boil to death amid the gradual temperature change.
Company reputations work much the same way, says Peter Cipollone, director, product management, text mining and visualization at Factiva, a Dow Jones & Reuters company. A reputation may change slowly over time in one direction or another, but if a company doesn’t catch a negative swing early on, it might one day find itself up to its neck in boiling water. The good news is there are now text mining tools that enable organizations to sift through the reams of unstructured data on the Web and spot reputation trends long before they become real problems.
Take the example of Maytag, says Cipollone in his white paper, “Boiled Frogs and Your Organization’s Reputation.” Early in 2002 the company was enjoying strong profits and a climbing share price. Executives were missing the growing undercurrent of frustration with Maytag’s Neptune washer, however. The product, which rolled out in 1997, was suffering from control boards shorting out and door leaks that led to mold. Ultimately, the problem resulted in a $33.5 million class action lawsuit and the destruction of Maytag’s reputation.
What company executives didn’t know in 2002 was that the Neptune issue had been playing out slowly on the Web for several years. “Consumers repeatedly reported Neptune’s shortcomings on blogs,” says Cipollone. “These messages about declining quality control were growing in the nooks and crannies of the Web, though the mainstream press hadn’t begun widespread reporting on the issue. Because the data was looked at one snapshot at a time, it was easy to miss the growing concerns around the quality control issue.”
If text mining tools had been used to chart the historical changes in quality control comments, says Cipollone, it would have been clear instantly that quality was an issue of increasing concern. In fact, when looking at the Maytag problem Factiva’s new Factiva Insight: Reputation Intelligence tool found a change developing over time and detected the signals that alerted other appliance manufacturers to this important competitive issue.
Unlike data mining, which extracts information from highly structured databases, text mining extracts meaning from unstructured data such as Websites, blog posts or stories from newspapers, magazines and wire services. “In text mining, patterns such as word proximity and sentence structure identify additional meaning from text and extract important information for further analysis,” says Cipollone.
Text mining capabilities, he adds, enable companies to “uncover the business issues and social trends affecting corporate reputations now and in the future.”
For more information, visit www.factiva.com.