Lenovo was near final design on an update to the keyboard layout of one of its most popular PCs when it spotted a small, but significant, online community of gamers who are passionately supportive of the current keyboard design. Changing the design may have led to a mass revolt of a large segment of Lenovo’s customer base—freelance developers and gamers. The Corporate Analytics unit was using SAS as part of a perceptual quality project. Crawling the Web, sifting through text data for Lenovo mentions, the analysis unearthed a previously unknown forum, where an existing customer had written a glowing six-page review of the current design, especially the keyboard. The review attracted 2,000 comments! “It wasn’t something we would have found in traditional preproduction design reviews,” says Mohammed Chaara, Director of Customer Insight & VOC Analytics. It was the kind of discovery that solidified Lenovo’s commitment to the Lenovo Early Detection (LED) system, and the work of Chaara and his corporate analytics team. Lenovo, the largest global manufacturer of PCs and tablets, didn’t set out to gauge sentiment around obscure bloggers or discover new forums. The company wanted to inform quality, product development, and product innovation by studying data—its own and that from outside the four walls. “We’re mainly focused on supply chain optimization, crosssell/up-sell opportunities and pricing and packaging of services. Any improvements we make in these areas are based on listening to the customer,” Chaara says. SAS provides the framework to “manage the crazy amount of data” that is generated. The project’s success has traveled like wildfire within the organization. Lenovo initially planned on about 15 users, but word of mouth has led to 300 users signing up to log in to the LED dashboard for a visual presentation on customer sentiment, warranty, and call center analysis. The Results Have Been Impressive • Over 50% reduction in issue detection time. • 10 to 15% reduction in warranty costs from out-of-norm defects. • 30 to 50% reduction in general information calls to the contact center. Looking at the Big Picture Traditional methods of gauging sentiment and understanding quality have built-in weaknesses and time lags: • Customer surveys only surface information from customers who are willing to fill them out. • Warranty information often comes in months after delivery of the new product. • It can be difficult to decipher myriad causes of customer discontent and product issues. In addition, Lenovo sells its product packaged with software it doesn’t produce, and customers use a variety of accessories (docking stations and mouse devices) that might or might not be Lenovo products. To compound the issue, the company operates in 165 countries and supports more than 30 languages, so the manual methods to evaluate the commentary were inconsistent, took too much time, and couldn’t scale to the volumes of feedback it was seeing in social media. The sentiment analysis needed to be able to sense nuances within the native languages. (For example, Australians describe things differently than Americans.) The analysis-driven discovery of an issue with docking stations provided the second big win for Lenovo’s LED initiative. Customers were calling tech support to say they were having issues with the screen, or the machine shutting down abruptly, or the battery wasn’t charging. Similar accounts were turning up on social media posts. Sometimes, though not always, the customer mentioned docking. It wasn’t until Lenovo used SAS to analyze the combination of call center notes and social media posts that the word docking was connected to the problem, helping quality engineers figure out the root cause and issue a software update. “We were able to pick up that feedback within weeks. It used to take 60 to 90 days because we had to wait for the reports to come back from the field,” Chaara says. Now it takes just 15 to 30 days. That reduction in detection time has driven a 10 to 15% reduction in warranty costs for those issues. As warranty claims cost the company about $1.2 billion yearly, this is a significant savings. Although the call center information was crucial, the social media component was what sealed the deal. “With Twitter and Facebook, people described what they were doing at that minute, “I docked the machine and X happened.’ It’s raw, unbiased and so powerful,” Chaara says. An unforeseen insight was found when analyzing what customers were saying as they got their PCs up and running. Lenovo realized its documentation to explain its products, warranties, and the like was unclear. “There is a cost to every call center call. With the improved documentation, we’ve seen a 30 to 50% reduction in calls coming in for general information,” Chaara said. Winning Praise beyond the Frontlines The project has been so successful that Chaara demoed it for the CEO. The goal is to configure a dashboard view for the C-suite. “That’s the level of thinking from our senior executives. They believe in this,” Chaara says. In addition, Chaara’s group will be formally measuring the success of the effort and expanding it to measure issues like customer experience when buying a Lenovo product. “The application of analytics has ultimately led us to a more holistic understanding of the concept of quality. Quality isn’t just a PC working correctly. It’s people knowing how to use it, getting quick and accurate help from the company, getting the non-Lenovo components to work well with the hardware, and understanding what the customers like about the existing product—rather than just redesigning it because product designers think it’s the right thing to do. “SAS has allowed us to get a definition of quality from the view of the customer,” Chaara says.
Questions for Discussion
1. How did Lenovo use text analytics and text mining to improve quality and design of their products and ultimately improve customer satisfaction?
2. What were the challenges, the proposed solution, and the obtained results?