Listening to the Voices of Customers

Traditional CRM analysis exposes only 20 percent of the valuable customer insight that your company captures today. Enterprises have invested heavily in CRM solutions that collect tremendous amounts of information on their customers during product-support, service-request, or transaction-processing interactions. During these exchanges, a wealth of information is captured in call center notes, or verbatims, such as the customer's emotional state, sense of urgency or impatience, competitive insights, and specific and complex issues associated with buying, using, or configuring a product or service.

But it is only the structured data--the information that can be easily stored in the rows and columns of traditional databases--that is typically analyzed and mined. In many organizations today the unstructured information, such as call center verbatims, is reviewed only occasionally--usually one document at a time--and rarely used to identify trends, root causes of problems, or opportunities for improvement. In fact, there is even more unstructured information residing outside the enterprise. Web forums have gained in popularity, and customers are sharing their grievances and expectations through online product reviews, blogs, and social networking sites. This consumer-generated content is growing rapidly in both volume and value.

The complete range of unstructured information--gleaned from both inside and outside the enterprise--must be thoroughly analyzed if a company is to acquire a complete understanding of its competitive threats and how to improve products and services to retain old and acquire new customers. Traditional CRM analytics that leverage only structured data can tell a company how many customers called, average call durations, specific transaction amounts or products purchased during an exchange--and, perhaps, the broad category of issue discussed during a call. They cannot, however, answer the important questions that help a company improve product experiences, assess quality-of-service issues, or proactively identify a competitive trend that might impact future corporate performance.

Text mining enables a company to leverage unstructured information and answer the following questions:

What's behind the trend in three- to five-year customers canceling their contracts despite steady prices?

A new product that sold well after a marketing launch appears to be falling off precipitously, and returns are much higher than expected. Why? Product complexity? Quality? Installation issues?

Another company just launched a directly competitive product that is impacting your sales. What are your customers saying about the competitor's product?

Anecdotally, you believe that angry/frustrated calls across a national chain of retailers are increasingly associated with product installations. Is the increase quantitatively measurable? What's causing the increase?

You have a sense that a product-support issue that is hard to categorize is becoming more common, but the information is being captured only in call center notes. Is the issue truly increasing in frequency? Is your staff able to resolve the issue?

What's the must-have toy going to be for Christmas? How do customers expect to buy the product?

Your competitor's drug has just been pulled from the market because of safety concerns. How have perceptions of your product quality and safety been impacted? Is a campaign necessary to calm fears of your customer base?

Both structured data (which quantifies historical customer transactions) and unstructured textual information (which reveals customer feelings and desires) can provide meaningful insights that drive a variety of sales, marketing, and customer-support decisions. Early adopters of text mining--including companies in consumer goods, retail, life sciences, and financial services--are synthesizing all of the structured data and unstructured text available to them to dramatically enhance decision-making.

Commercialized text-mining solutions--the product of simultaneous development across BI, data warehousing, CRM, and natural language processing technologies--integrate seamlessly with companies' traditional reporting tools. Familiar techniques, such as scorecarding and creating slice-and-dice and drill-down reports, can be applied for the first time to unstructured information, and new techniques of sentiment analysis and automatic categorization can be applied against customer information. With a holistic view of internal information assets and external, consumer-generated content, companies can use commercialized text mining to answer any question of any internal or external data source, using any analytical tool. Text mining ultimately makes a company more agile and adaptive, capable of adjusting course at the onset of trends.

Enterprises have never had available to them so many sources of information or such sophisticated tools for discerning the true sentiment of the market. Commercialized text mining tunes a company into its customers.

About the Author

Sid Banerjee (sid.banerjee@clarabridge.com) is the cofounder and chief executive officer of Clarabridge, a provider of text mining software used by many Global 1000 companies to improve customer experience management (CEM). Banerjee has amassed more than 15 years of business intelligence leadership experience and is a highly respected thought leader within the text mining community.

Related Articles

Radian6 Gets Sentimental

The social media monitoring company announces new sentiment analysis technology for its customers.

Pilgrim Software Synchronizes Complaint Logging

The newest offering from the provider of enterprise compliance and quality management solutions takes the human factor out of unifying data.