Macys.com has made a name for itself by analyzing consumer data to gain insight into its loyal shopper base. The retailer, which features more than 60,000 products online and leverages multiple digital channels to share product information cross-channel, has armed itself with a set of internal cloud-based, high-performance analytics tools designed to improve customer segmentation, market-basket analysis and real-time personalization cross-channel, based on a broader scope of product attributes.
“We need to understand where the demand is coming from and how to use the data to know how to best serve our shoppers,” said Kerem Tomak, VP marketing analytics, macys.com. “But managing the volume of data is an increasing challenge.” Indeed, he described the amount of data that the retailer collects as “mind-blowing.”
It is made even more so by the fact that macys.com is looking to harness three dimensions of data: volume of the data, variety of data (including structured and unstructured information), and the velocity with which they receive and process it.
“We have a short time frame to process these large product attributes and match these to customer needs,” Tomak explained. “Traditional BI or near-real-time solutions don’t meet our needs because we have less than a minute to really > capture customer attention. If we don’t do it the right way, she’s gone.”
Knowing how crucial it is to quickly turn raw data into actionable information, macys.com turned its attention to a high-performance analytics tool from SAS, Cary, N.C. The retailer linked the SAS Enterprise Miner to a Hadoop cluster, an open-source, data-management platform initiated by Yahoo and Google. This private cloud platform sits behind macys.com’s firewalls and processes large data sets in a distributed computing environment.
Macys.com compiled two years worth of data, which ranges from 20 terabytes to 30 terabytes. As new data comes into macys.com servers, it is stored on Hadoop and passed to the SAS tool, where macys.com analysts compare information with its historical database