Case Study: Buy.com
Company
In an approach that Buy.com chief operating officer Brent Rusick describes as “more efficient than any other out there,” the e-tailer's orders are filled directly from its distribution partners. Unlike some of its competitors, the site has no warehouses and so avoids the expense of inventory.
Buy.com fills each order – it sells computers, software, electronics, games, books and movies – by choosing among one of several wholesalers or distributors that stock that item. The distributor with the best combination of price, availability, and proximity to the customer makes the sale.
Problem
While this non-inventory approach is something brick and mortar retailers only dream of, in actual practice it presents Buy.com with a formidable logistical challenge.
According to Rusick, the company has pared down its staff to about 100 employees. So this relatively small staff maintains a site that Rusick says averages about 2,000 simultaneous shoppers per minute, who choose among 1.5 million product SKUs; the site's customer base is 5 million, and at times it handles close to 100,000 transactions in an afternoon. And, to top it off, Buy.com recently announced a free shipping policy for all orders under 120 pounds (which some industry analysts say is a response to Amazon.com's policy of free shipping on orders over $49).
So, given the site's volume and reduced staff, its challenge is getting products from wholesalers to consumers, quickly and accurately, in a cost-effective manner.
Solution
Buy.com handles this challenge with an operation that is essentially software driven. Its e-commerce database is powered by Microsoft SQL Server, with Microsoft Commerce 2000 as the platform's core. But Rusick points out that, while the Microsoft software is the engine's basis, “We've taken it and enhanced it,” continually modifying it for several years. “Some people think you can take that product and just plug it in and all of sudden you're got an e-commerce store. That's not it – there's a lot of customization and development that goes into building a platform – literally hundreds of millions of dollars,” he says.
Rusick says the site is designed to handle 10 times the average transaction volume, and on a few occasions in which it offered several promotions simultaneously, it has come close to needing this capacity.
Running alongside the Microsoft software is E.piphany, an application that tracks customer purchases so Buy.com can make logical recommendations at shoppers' return visit. E.piphany also makes purchase recommendations to shoppers based on customers with similar attributes, as well as providing data for e-mail campaigns.
As Rusick describes it, Buy.com's operation is divided into a few different aspects. The front end, “the basket,” is the consumer interface – standard in online retailing – that handles shopping processes. It verifies credit cards, adds items to the shopping cart, and routes orders to the right channel (electronics, music, etc.).
But the most highly refined aspect of Buy.com's operation is its fulfillment engine, which determines the best source to fill each order. “On a batched basis every few minutes, we send orders to our suppliers. They in turn send back a confirmation that yes, we've received X number of orders and we're filling these today and we're filling those tomorrow, and this communication goes back and forth on a continual basis each day,” Rusick says.
The software that drives this process operates by “a set of well-defined business rules that define how orders get filled,” he says. Since most of the site's items could come from multiple suppliers, when a shopper buys, for example, a DVD player, these rules compute the myriad variables of price-availability-location to determine which DVD wholesaler makes the sale. These rules are the result of “a lot of hours of experience and trial and error,” Rusick says.
“There are some rules that are rock solid, but as the environment changes, be it the economy or suppliers, other rules are in a state of flux,” he says. “You add free shipping into the mix and then you change the set of rules again. You say, okay, now this particular attribute of a supplier is not as relevant as it was two weeks ago because we've made other business decisions.”
“That's really the trick,” he adds. “You've got all these suppliers and you're looking into their real time inventory, and our SKU count could vary by a couple thousand SKUs or more on a daily basis, due to all the various suppliers and the categories we're in. So maintaining the database of products is a continually evolving dynamic process.”
As the software drives the fulfillment engine, it simultaneously sets prices based partially on what Buy.com's competitors are offering. “We use a crawler, and it goes out and checks pricing, then we update pricing on a daily basis,” says Rusick. For instance, the Buy.com crawler brings back continuous data on what Amazon.com is charging, adjusting Buy.com prices accordingly. Since Buy.com recently announced it will beat Amazon's book prices by ten percent, this software plays a key role. For each of Buy.com's channels – books, computers, electronics – the company has different target sites that it crawls to track its competitors.
As efficient as Buy.com's model appears to be, it has had its share of twists and turns while building its Internet brand. After a successful IPO in 2000, its share price dwindled with the Internet downturn, and in 2001 Buy.com CEO and founder Scott Blum took the unusual step of buying the company back, taking it private again. He also slashed staff, hence the company's roster of approximately 100. Even with its leaner profile, Jupiter Media Metrix analyst Ken Cassar says the company can not afford its free shipping offer. “In order for Buy.com to make any money off of this, their order size has to be two units per order. There's no incentive in the offer to drive the order that high,” Cassar says.
Yet Rusick is upbeat about the offer's results. “It's been very effective for us. It's a great promotion that's going to continue on to the foreseeable future,” he says. “And we think that because we have such a highly efficient model we'll be able to pass savings on to the consumer for a long time to come.”
Reprinted from ECommerce Guide.
