RealTime IT News

Tracing Back Unhappy Customers Online

NEW YORK -- Companies need to know about and respond to criticism on the Internet, said Matthew Brown, Forrester Research analyst at the Enterprise Search Summit in his speech on connected structured and unstructured information.

He pointed to a complaint about BMW posted in 2007. He said that companies lack the analytical tools to find customers who are complaining about their products.

"The goal is to better understand customers," he said. He added that the tools are needed. "Our studies show that bad news travels very quickly today," he said.

How fast does bad news travel? Brown pointed to an incident last year when a Florida news site re-posted a story from 2002 about United Airlines going bankrupt. The story was picked up by Google as breaking news and United Airline's stock (NASDAQ: UAUA) fell by 75 percent in fifteen minutes. Executives at the company scrambled to find the cause.

"We need discovery-centric approaches to business intelligence," said Brown.

He said that traditional business intelligence systems are designed for strategy modeling. "We think about the questions the system will be asked and expect customers to ask those questions, such as what was last year's revenues."

Traditionally, he said, companies dump all of their databases into one file, so notes would end up in one cell and become virtually useless. "There's no understanding of the meaning of the data, or linkages to other systems," Brown said.

He added that a company called Attivio has a new approach that builds a database consisting of metadata that preserves the relationships between data points and enables better searches.

Learning from Google

Newer approaches to business intelligence borrow the user interface of search, Brown said. "Cognos and Business Objects were typically used by power users. Search gives us the possibility of a white box query."

Both search and business intelligence tools are improving in a converged manner he said. Search is getting better at parsing structured data and business intelligence tools, which were always good at structured data, are working on search's strengths, such as handling taxonomies.

His comments came a day after SAP unveiled a new natural-language front end search for its business analytics software, demonstrating the convergence that Brown was talking about.

His comments came in the session after Nitin Mangtani, lead product manager for Google enterprise search explained at the conference that Google's search box is actually more complex than it appears and that power users of the search box can do more with it than regular users.

Brown noted that Google's OneBox appliance for business can, similarly, do more for power users. "A particular string of keywords, 'sales 2009,' can trigger a query to Cognos, for example," he said.

Google is also working on making the appliance easier to use for regular uses. "It can now take a natural language query and turn it into a SQL query," he said.

The entire Internet is an enterprise database

In case all of this sounded too vague, Brown provided a specific case. He said that business intelligence provider Clarabridge had an unnamed customer in the hotel industry that imported customer reviews of the hotel experience from public Web sites and learned some very specific lessons about what made the customer experience good -- and bad.

"They learned that if customers find even one hair in the bathroom, they won't return," Brown said.

In order to extract that data point, they pulled out customer reviews with specific words (i.e, "hair" correlated with "bathroom") and looked at the numbers attached to those words, numbers indicating levels of satisfaction.

Brown added that companies that are masters of this technology use it not only to judge their own performance, they scour the Web to track their competition's pricing and performance.

But few companies are actually masters of the tools that have been made possible by the convergence of business intelligence and search. "The biggest inhibitor is not a technology challenge," Brown said. "Often, the BI employees and the search employees don't know each other. They're in different reporting structures."

The convergence between search and business intelligence has therefore only just begun, Brown concluded. It will accelerate as the software providers adapt to the convergence. "We expect acquisitions," he said. "A number have happened already."