Calculating text analytics ROI: Start small and focus on customer data

Calculating the return on investment (ROI) and building a business case for text analytics to analyze unstructured content like emails, PDFs and blogs is challenging but worthwhile, as Intuit found.

For Intuit Inc., maker of the popular TurboTax software, customer feedback that was once hidden amidst the flood of customer emails, comments and forum responses ultimately saved the company millions of dollars once it implemented text analytics.

Like many, the Mountain View, Calif.-based company solicits customer feedback via email, surveys and forums. Responses often return in the form of unstructured content -- freeform text, comments detailing customers' likes and dislikes. Rather than hiring an army of analysts to read each comment, Intuit uses text analytics software from Reston, Va.-based Clarabridge to gain a broad view of its unstructured content, to analyze customer sentiment and to identify trends.

"Once we understand what motivates our customers, we can make improvements," Chris Jones, Intuit's manager of analytics, said earlier this week at the Text Analytics Summit in Boston. "Feedback is a gift."

Experts estimate 80% of corporate data takes the form of unstructured content. Until recently, most organizations had no efficient way to summarize and analyze it -- the way business intelligence (BI) technology does for structured content. However, in the last several years, text analytics technology has developed to the point where analyzing unstructured data is a much easier proposition. But the technology is expensive, and calculating its return on investment (ROI) is a difficult task, according to experts.

Companies seeking to establish the ROI of a text analytics project should start their deployment with a specific, targeted set of unstructured data to prove the business case, suggests Dave Kasabian, research director with Boston-based AMR Research.

"ROI is difficult to calculate, but I think one of the areas that's most measurable from an unstructured perspective would be customer data, because customer retention and satisfaction are key revenue drivers," Kasabian said. "Being able to leverage unstructured data to analyze call logs, emails and even in some instances [call transcripts] gives you the ability to get more insight into customer sentiment and to reduce customer churn."

By analyzing the text in customer feedback forms, Intuit was able to determine that customers were less than satisfied with its website functionality. Some adjustments to the navigation window resulted in a 50% increase in user self-service. As a result, call volume to Intuit's support call center shrank, saving the company millions. In this case, the ROI was clear: Intuit's investment paid for itself in less than a year.

Unstructured customer data just a start

Once the ROI of applying text analytics technology to unstructured customer content is proven, it is easier to convince other parts of the organization of its value. Indeed, at most organizations, unstructured content of all forms abounds, according to Bill Inmon, author and data management industry analyst.

Corporate contracts, for example, are full of valuable unstructured content, according to Inmon, who recently founded Forest Rim Technologies, a Littleton, Colo.-based firm specializing in unstructured data integration. With the naked eye, a lone executive can examine just one contract at a time, he said. With text analytics software, that same executive can analyze thousands of contracts simultaneously.

"Now you can look at 100,000 contracts at a time," Inmon said. "And you can ask questions such as, 'What is my total corporate liability according to my contracts?' or 'What contracts expire on January 1 next year?' " With those insights, executives can rewrite suspect or vague contract terms, for example, or develop new pricing models, potentially saving thousands of dollars.

Similarly, purchase orders, invoices, and product specifications also contain a wealth of unstructured content just waiting to be analyzed, Inmon and Kasabian agreed.

Focus on text analytics results, not the technology

The major software and business intelligence vendors have taken a keen interest in unstructured content recently. It was a hot topic at SAP's Sapphire conference earlier this year, Kasabian said, and Business Objects, since acquired by SAP, itself acquired text analytics firm Inxight Software last July.

At the Text Analytics Summit, industry analysts said they expect market consolidation to continue. Some, including Seth Grimes, principal consultant with Takoma Park, Md.-based Alta Plana Corp., speculated that text analytics may merge into the database or other applications like extract, transform, load as a result.

Not surprisingly, analyzing unstructured content across the enterprise is an expensive undertaking. If it weren't, the mega-vendors like SAP and IBM probably wouldn't be interested in it. Kasabian said companies should expect to spend as much analyzing unstructured content as they do analyzing structured data via traditional BI technology.

Selling such an expensive technology to management can be a challenge. In addition to focusing first on a specific unstructured data set for an easy win, Inmon also suggests touting the potential benefits of analyzing unstructured content -- like reduced customer churn -- rather than the technology itself.

"Trying to sell unstructured technology is bound to fail. But trying to sell the value proposition created by analyzing unstructured content is a no-brainer," Inmon said. "Don't sell the technology. Sell what the technology does for you."

Next Steps

Learn how text mining tools and prediction markets extract insight from unstructured content

Listen to this podcast on unlocking and integrating unstructured content

This Content Component encountered an error

Pro+

Features

Enjoy the benefits of Pro+ membership, learn more and join.

0 comments

Oldest 

Forgot Password?

No problem! Submit your e-mail address below. We'll send you an email containing your password.

Your password has been sent to:

-ADS BY GOOGLE

SearchDataManagement

SearchAWS

SearchContentManagement

SearchCRM

SearchOracle

SearchSAP

SearchSQLServer

Close