• March 21, 2015

Driving Competitive Edge Through Predictive Customer Support

Is it possible for the enterprise to anticipate customer complaints before they occur? Can organizations develop strategies to address product or service issues before they become costly problems? Where do warning signs of potentially crippling product issues first emerge?

The new competitive edge is not just knowing your customers or predicting their behaviors, but acting on those predictions before they reach a pain point. Welcome to the era of predictive customer support (PCS) where enterprises can both forecast reaction and drive solutions to predicted challenges in advance. I’m fortunate to be on the Hewlett Packard Enterprise front line of this customer support innovation.

PCS exposes product or service issues that were previously unknown. It does this by mining and analyzing social media, geographic information systems and other contextual data. In conventional enterprise environments, it may take weeks or months for new problem patterns to emerge. PCS reduces this process to weeks, days, or even hours.

PCS Intelligence

PCS empowers call centers and service desks to be more proactive. For example, customers are more likely to express sentiments on social media platforms before they contact an enterprise service center. By monitoring online platforms such as YouTube, Yahoo Answers, Twitter, Facebook, blogs, and forums that aggregate customer reviews, PCS enables the enterprise to anticipate emerging problems through heightened visibility.

PCS exposes what is currently not known. If we start with all available information and subtract what we already know, what is left contains the unknown. The challenge is that the unknown data is massive in volume and full of noise. So PCS deploys a series of filters to sift out subjects of interest, such as the makes and models of cars, smartphones, and laptops, as well as negative sentiments, geographies, and languages. Further, the specific references to the incidents are extracted using natural language processing.

Through a technology called Discrete Wavelet Transform we can determine what is significant and what is not. We may start with 1,000 incidents; narrow it down to 20, and identify the top five. Then we correlate the more significant incidents with contextual data such as demographic information, weather, climate, geographic information, and types of businesses. We’ll find out, for example, if a PC fan is failing, it’s probably occurring during summer in warmer states.

Driving Predictive Value

Once the information is contextualized for relevance, a report is issued to the call center. These reports detail upcoming unknown incidents for a given product and the conditions under which they occur. The call center can get ahead of issues by developing solutions such as upgraded training materials and updated knowledge bases for online and automated service systems. The same reports may be forwarded to product development groups, to address the root causes and potentially reduce warranty costs.

Result: PCS can significantly sharpen the enterprise’s competitive edge. It dramatically improves customer service and satisfaction while reducing the costly escalation of problems from first level service interactions. And it generates further business value as new useful information works its way into product development, manufacturing, and warranty processes.

The enterprise should never look at customer support the same way again. PCS is a breakthrough; a predictive force that can drive customer satisfaction, cost savings, business value, and competitive advantage across the bottom line.