September 01, 2011, 6:04 AM — Car buyers looking to purchase new or used vehicles have relied on the Kelley Blue Book to give them an estimate of how much a vehicle is really worth.
Until recently, the numbers were largely educated conjectures that Kelley analysts arrived at by running a handful of car sales and other metrics through a rudimentary pricing algorithm.
Not anymore. Kelley has begun using sophisticated data analytics tools to sift through volumes of historical and current data to arrive at what company executives say are far better car valuations.
"We went from using megabytes of data to using terabytes of data" for estimating car values, said Dan Ingle, vice president of analytic insights technology at Kelley Blue Book. "We moved from simply averaging data to running sophisticated models," based on transaction and regional data, he said.
KBB is an example of how a growing number of small and medium-size businesses (SMBs) are successfully applying data analytics to improve the way they do business. Driving the trend is the growing availability of relatively low cost, specialized tools for analyzing large sets of structured and unstructured data, according to an upcoming study by The Data Warehousing Institute .
Kelley is an early adopter in the SMB space. The company started embracing analytics technologies about three years ago in a bid to make better use of its historical data as well as the new data it had begun capturing from website users and social media, Ingle said.
The company found its Microsoft SQL-based business intelligence and data warehousing infrastructure was not scalable enough to handle Kelley's growing data analytics requirements.
About two years ago, the company moved over to a new IBM Netezza Twinfin data warehousing appliance which it supplemented with a second similar system last December. The two systems together, with software from Information Builders and MicroStrategy, form the core of Kelley's new data warehousing and business intelligence capabilities.
Kelley is also using a variety of predictive analytics, data mining and text analytics tools from SAS Institute to help analyze the data it collects. Much of the analytics used to deliver new and used-car values, targeted advertisements, customized offers and reviews on the company's website kbb.com are powered by SAS' software.
From being a legacy book-publishing company, Kelley has transformed into a completely analytics-driven company, said Shawn Hushman, Kelley's vice president of enterprise analytics. Analytics powers almost every facet of the company's business, including its car valuation processes, market research, customer analytics, financial forecasting and demand planning, he said. Kelley's analytics group has grown from one person to 23 in just just three years.
The effort has begun to pay off in several areas. One of the most dramatic improvements has been in Kelley's car valuations, which are based on a far richer data set than before, Hushman said. From updating car values once every five months, Kelley now updates values every week for more than 20,000 vehicles. "We could do it on a daily basis if we wanted to," he said.
Kelley's efforts to mine social media and web data such as weblogs and clickstream data has also greatly improved the company's ability to forecast ad inventory, gauge customer sentiment and predict user behavior.
"If you look at where we have focused our analytics efforts, it has been around two disparate data sets," Ingle said. One of them is about providing better vehicle valuation data, while the other is about the company's Web presence.
One of the more interesting aspects of Kelley's data analytics effort is its focus on social media and text analytics, said Tapan Patel, global marketing manager for SAS' predictive analytics and data mining group. It's rare for companies, especially of Kelley's size, to get started on social media and text analytics so early in their analytics life-cycles, he said.
Even many larger companies that have been doing data analytics for years are only now beginning to explore how to take advantage of social media and text-mining tools to boost their analytics capabilities, he said.
