Putting Big Data to Work

Dell masters the art of information to climb to the top, one byte at a time.
Big Data

Dell On the Move

Dell is a company in transition.

After dominating the enterprise PC markets for decades, the Texas-based configure-to-order manufacturer is making a definitive move away from the product side of the business and headlong into services and solutions.

Doing so has meant totally restructuring its business model, rethinking even the core configure-to-order option that helped make the company great.

"Historically, Dell has worked with low inventory, at best no inventory," Schmidt explains. "But in our new model, as we move forward over the next couple of years -- and actually coming into effect very quickly -- we are going to sell a significant amount more out of inventory."

This sounds like a manufacturing problem, or at least just a simple business problem. But for Dell, it hits squarely in the data.

Just a few years ago on the Dell website, variations in models, software configurations, memory, screens, design and every other customizable feature resulted in over seven septillion possible configurations of Dell products -- that's 7,000,000,000,000,
000,000,000,000 possibilities.

Moving to an inventory-based system required trimming a dozen or so zeroes off that range in such a way that the company could still satisfy the diverse needs of its broad customer base. And that meant putting Schmidt's data to work.

The four dimensions of big data

To do that, Schmidt and his team, led by Bart Crider, IT director for enterprise BI at Dell, created a new system called "optimized configuration" to help transform the product model.

"The analytical team clustered high-selling configurations from historian order data to create technology roadmaps," Schmidt explains. "They created automated algorithms to identify potential demand coverage for specific configurations -- so the data could tell us what configurations we should be building and what configurations we wanted to put into inventory."

"Basically," Crider adds, "we were able to take all of the historical data from our Teradata warehouse, and run cluster analysis to determine the most common configurations our customers were choosing."

Going through that pile, he says, they were able to trim those seven septillion options down to a couple million, even identifying certain models so common that the company could stock into a preconfigured inventory to sit, ready to ship. 

"We targeted the ones that people were most interested in and which they could build within the best margins," Crider says. "We set it up in a system so that, if you order today, I can have it to you tomorrow."

Such a system is dangerous in a typical setting. In the wildly shifting, tumultuous PC market, a warehouse of product is too commonly a warehouse of shrinking value -- unless you have another warehouse of data on the other side backing you up.

But of course the real test is in the results. And, according to Schmidt, this "optimized configuration" experiment passed with an easy A, bringing the company an extra $40 million in positive revenue.

"From a Dell perspective, we fully agree that the rise of big data has a massive amount of value," Schmidt says. "I think a couple of years down the road, we're going to be able to drive faster insights and more value on data to information than we ever imagined possible."

"We're pretty excited about it."

Please or Register to post comments.

Subscribe to IW Newsletters

IW Marketplace - Buy a Link Now