IBM has introduced a new service designed to help organizations mine data from various sources, including audio recordings, call transcripts, emails, survey results and demographic data, and then analyze that data to better gauge customer sentiment and improve marketing and customer service campaigns. Called IBM Voice of the Customer Analytics (VOCA) and developed by IBM researchers, the service is an offering within IBM's Managed Business Process Outsourcing (MBPO) group. It is available now as a standalone, subscription-based service, but IBM also intends to embed the technology in its MBPO customer-relationship management offerings.
Central to the offering is IBM-developed data mining technology that's the core of IBM Content Analyzer, which is able to comb through both structured and unstructured data culled from a variety of different sources, include data in a CRM repository, audio content from call logs and even documents of customer surveys. But IBM researchers spent the last year adding more advanced analytical tools and algorithms on top of IBM Content Analyzer, says Kevin English, global offering lead for CRM Analytics, IBM MBPO. One function is the ability to examine text and look for context to determine whether the text--which for example could be from a conversation with a customer via a phone call--is positive or negative in its sentiment. Another function, which leverages clustering technology, allows for ability to find patterns within the data in order to determine a trend or make more granular observations. The new technology also can classify the data in order to separate out, for example, audio recordings that were inaudible. "This is a good way to eliminate some of the manual work that is often required when examining data," says English. "This is a compliance and quality tool."
A global manufacturing company headquartered in the U.S., which asked not to be identified, is using VOCA to help improve interactions with customers, according to Larry Bates, a Voice of the Customer manager and innovation lead who works on site at the manufacturer as part of IBM's Customer Care group. The company began a pilot in the second half of 2008 and started using the service in the beginning of 2009. "Our aim is to understand customer interactions better and improve the overall experience. To do that, we're using VOCA to identify the drivers of customer satisfaction and dissatisfaction when they interact with us to identify ways that we can improve our processes, policies and execution. We also use VOCA to look for ways to improve efficiency, but this is secondary to improving the customer experience," he says. The manufacturer is mining and analyzing data using VOCA from its CRM system, from the marketing organization and from a third-party organization that fields customer satisfaction surveys. The company is also starting to pull in some data from social media sites for analysis.
Bates says VOCA is helping the manufacturer predict--almost in real time--interactions that are most likely to leave the customer dissastified. Three times a week, Bates and his team upload all open customer services cases to VOCA, and VOCA spits out a prioritized list for proactive mediation. VOCA analyzes 120 different call attributes, such as the use of specific words from agent notes, the amount of time it takes to resolve the case and the number of calls it takes to resolve, to predict the outcome. "Roughly half of our customer interactions involve opening an issue resolution ticket. We predict which cases look like they will be resolved unsatisfactorily for preemptive interaction," he says. "The result has been that we are closing those cases with a much higher satisfaction rate than before, with a 50 percent improvement rate for these cases." VOCA is also helping the manufacturer conduct periodic ad hoc analysis, including searches for correlations between interaction attributes and subsequent purchases of products to predict future purchases. The manufacturer has contracted with IBM to use the VOCA service through 2010. Future plans include analyzing more content from social media and transcripts from live Web chats with customers. "We also plan to start analyzing comments about our partners and distribution network," Bates says.
IBM has also been testing the service with an electronics company that's using VOCA, in particular the clustering technology, to better understand why customers are calling the tech support and product support call centers. Using the clustering capability, the company can isolate by product the top issues customers are calling about, says English. Another early VOCA customer is a telecommunications company that's using VOCA to help segment its clients not just by standard demographic attributes, but by past experiences with the telecommunications company.