Time to listen -- text mining needs greater adoption by customer service departments.
Posted Jul 1, 2007
Drawing from my personal experience, when I write to customer service it more often than not feels like my email goes into a black hole. I wonder is there someone on the other end who immediately hits the delete button when my email appears. Or is it collected along with the other thousands of emails to get to "one of these days"? As a customer who has strong brand loyalty and truly wants my favorite companies to improve and excel, why is it that they don't listen to me--the customer? Or is that they are listening, but no one is replying to me and letting me know?
We conducted an informal research study interviewing customer executives ranging from midsize companies to Fortune 100, and the analysis showed us that although all customer service executives recognize the importance of reading and responding to customer inquiries, they don't have the time to read through the tens of thousands of inbound email or web forum inquiries each day. What we discovered was, most of these companies have hourly paid workers who read on average 10 percent of the incoming mail and then have to categorize (or tag) the inquiry from a scrolling list of 50 to, in some cases, several hundred options, to match the exact complaint.
What's astonishing is that during the interview process all of these companies acknowledged their current customer inbound response process and turnaround time is far from optimal. What's even more amazing is that there is technology available today that can automate and improve the customer service response time and accuracy of resolution by using text mining. Text mining is the process of collecting text (referred to as unstructured data), extracting out key terms and concepts, appropriately routing the inquiry, and summarizing the voice of the customer.
Why aren't text mining solutions being used by every customer service center?
Although over the past few years there has been a growing buzz about text mining and what to do about the staggering amounts of data generated from daily emails, service requests, blogs, and customer forums, very few customer service departments have taken advantage of what the customer is trying to tell them.
Current processes of manually reading customer inquiries have become a futile effort due to the sheer mass of data being generated. A study conducted by the School of Information Management at the University of California at Berkeley revealed that instant messaging generates around 5 billion messages a day, or 274 terabytes of data a year, while email adds another 400,000 terabytes annually. Merrill Lynch has reported that more than 85 percent of the information within an enterprise is unstructured, and crucial customer service complaints are now extending beyond the corporate firewall when inquiries are not being responded to, as customers recognize their voices can be heard through peer-to-peer blogs.
Given the benefits that text mining solutions provide such as automating the tagging of inquiries, pulling out key complaints, and matching the appropriate correspondence back to the customer, why have customer service departments not embraced text mining solutions? First, it is extremely challenging to find one vendor that offers an end-to-end solution. To date, vendors have focused on niche areas of the text mining process, from term extraction to building domain-specific taxonomies to visualization of conceptual relationships in text. Software solutions that are marketed as licensed text mining suites have failed to deliver on scalability and require customers to have in-house text mining expertise. Along with unmet expectations, license price-points can vary by as much as 100 percent, contributing to a muddled market that makes potential customers leery and slows the emergence of new providers.
What criteria should you use when evaluating text mining solutions?
First and foremost, the highest rate of success is achieved through using text mining solutions delivered as a service. The main reason is that although text is very easy to manually read through one piece at a time, it requires several areas of expertise and technology to automate the process. Text mining the correct way is a combination of three skills: information retrieval (e.g., search), linguistics, and data mining.
When evaluating a text mining solution make sure it has the following functionalities:
It's delivered as a service from trained text mining professionals
The taxonomies are domain specific--if you're an airline company you don't want the same taxonomy that is used for a retail company.
The ability to provide use-clustering techniques to identify new trends, anomalies, and correlations in the entire customer base.
Ensure that the solution can handle not only your inquiries today but also support future growth. The amount of text data generated is only going to grow as it becomes cheaper to store data and as customers become more and more vocal.
The ability to have the solution up and running, the ability to process customer inquiries as they come in, and the ability to identify the appropriate routing and case management.
Everyone understands that it is more expensive to acquire a new customer than keep an existing customer. On top of that, brand loyalty continues to suffer as customers today more than ever have increased purchasing options. The customer is tired of not being heard. The customer service departments that implement text mining solutions will have competitive advantage, will get fewer complaints.
About the Author
Jeff Kaplan is principal at Apollo Data Technologies. Please visit www.apollodatatech.com.
Sponsored By: Jacada, Avaya, Confirmit, inMoment and BoldChat
Sponsored By: Genesys, Avaya, Verint, and Aspect
Sponsored By: Informatica
Sponsored By: Microsoft
Transform business process automation and people productivity with Microsoft Dynamics 365 and the Microsoft Cloud
New Gartner Report from Microsoft
Sponsored By: Microsoft
Sponsored By: Microsoft