There’s no better way to handle product quality problems than to prevent them from happening in the first place. That’s just what IT giant EMC intends to do and it recently launched an ambitious Big Data predictive analytics project to do it.
To understand the customer experience, EMC regularly survey’s its customers, talks with customer executives and benchmark’s its results against peers in the industry, according to Jim Bampos, Vice President of Quality at the Hopkinton, Mass.-based company. This allows EMC to identify and respond to customer pain-points, but only after they’ve become noticeable to the customer.
Bampos told me EMC wants to expand how and where it measures the customer experience in order to proactively identify product quality issues and fix them before the customer even realizes there’s a problem. The upshot could be worth millions of dollars or more in savings to EMC, but just as importantly it should significantly improve customer satisfaction and increase customer loyalty.
The good news is that EMC collects all the data necessary to perform predicative analytics already. Its storage, virtualization and other hardware-related products create torrents of machine-generated data – to the tune of 1.9 billion customer records per quarter –indicating how they are performing at any given time. The bad news is this data was strewn throughout the organization, living in isolated desktops and databases dispersed among various business units.
So the first step in creating a Business-Intelligence-as-a-Service platform, as EMC is calling its initiative, is to break down those data silos and create a single version of the truth related to its machine-generated data. While most organizations aren’t so lucky, EMC boasts an industry leading data warehousing unit in its Greenplum division, which it acquired in 2010, that is spearheading the job.
Starting around eight months ago, the Greenplum team set out to identify all the company’s machine-generated data sources, normalize the data into a common format, build the required data models, and store and process all that data in a Greenplum analytic database. EMC also tapped partners SAS Institute for performing predictive analytics and SAP to help with automating reports, Bampos said.
While the effort is ongoing, the Greenplum team has already automated the process of generating some reports from all this machine data, which now takes significantly less time and manpower than previously. A report that once took a developer six days to pull together is now automatically generated in eight minutes, Bampos said. Ultimately, the new predictive analytic capabilities will allow EMC to identify performance issues at the first sign of a problem, meaning intervention can take place before a costly replacement or other complex solution is required. The goal, in fact, is to forecast hardware failure rates two years in advance, Bampos said.
EMC also plans to eventually extend the project to its software products, correlate customer complaints with quality metrics, and help its own customers learn from EMC’s internal experience tracking and predicting quality assurance.
Gone are the days when just responding to customer complaints and quality issues after the fact was sufficient. Today, we all expect a seamless customer experience from the vendors we do business with and most of us are all too eager to share our displeasure to the world via Twitter and other means when our expectations aren’t met. This means vendors of all types, from IT giants like EMC to locally-owned small businesses, must consider new approaches to customer service that include leveraging Big Data to identify pain points before they reach the customers’ line of sight.
This requires investment in new technologies and services, but also a change in mind-set. The old ways of doing business must be replaced with an ethos of data-driven decision-making. To speed this process along, executive buy-in is key. At EMC, for example, President and COO Pat Gelsinger is an ardent backer of the Big Data predictive analytics project, which lessened any organizational resistance that might otherwise have slowed the project down, Bampos told me. Organizations undertaking Big Data initiatives likewise should look for support from the top.