To build or to buy?
June 2, 2015 10:04 AM
THE COMPANY: FragranceNet.com is a web-only retailer of perfume, cosmetics and related beauty products that’s been around the technology block a few times in the 18 years since it began selling online. Sales have steadily grown in line with overall industry sales growth: The e-retailer’s sales grew at a 14.6% compound annual growth rate over the last five years. With $190 million in 2014 sales, FragranceNet.com is among the top 10 health and beauty e-retailers in North American e-retailing, according to the Internet Retailer 2015 Top 500 Guide.
THE PROBLEM: The off-the-shelf technology the retailer used to run its order management system couldn’t keep up with the company’s growth, says Jason Apfel, FragranceNet’s president. “As our sales volume grew, performance of all aspects of our off-the-shelf system became sluggish,” he says. “Orders would take minutes to come up on the screen and that caused frustrations with our customer service representatives. Printing orders began to take hours and that caused warehouse delays.”
And if the retailer wanted to tweak the technology, it was at the mercy of its vendor—not only did the retailer have to conform to its timetable, it also had to hope that there would be enough demand from other clients to justify the desired change. For example, Apfel says FragranceNet wanted to enhance its coupon redemption process and the vendor was unable to provide the changes.
The problem raised by not being able to quickly modify the system to meet its needs became especially acute during the busy holiday shopping season, during which FragranceNet earns 40% of its annual revenue, Apfel says. Not only did the retailer want to move a greater quantity of orders through its systems and New York warehouse faster, it further sought a more efficient—and less error-prone—picking, packing and sorting process, especially because it uses a greater number of temporary workers during those frantic weeks. So in 2010, when faced with the decision of whether to invest more money into more vendor order management software or to build what it knew it needed, FragranceNet decided to build its own order management and fulfillment technology systems, a process that continues today and has extended beyond the warehouse.
THE SOLUTION: FragranceNet.com spent about 16 months creating its own software, hiring five developers for the task and related work. The result was the FragranceNet.com Order Activity Manager, or FOAM as Apfel calls it, which went into service in 2011. One key was to have a single system that handled a variety of important tasks. The new order management system has a homegrown fraud review system—Apfel says that out of some 1 million orders processed via the system, only 1,237 chargebacks have occurred, with a chargeback ratio of 0.11% of total sales.
But the order management software program wasn’t the only fulfillment-improvement project the retailer embarked upon. After that task came the building of the company’s own warehouse order-picking system, which put digital scanners in the hands of all warehouse workers. They are used to guide all activities, Apfel says, including picking and checking of outbound orders, stock replenishment and receiving.
“The main dashboard shows all the orders picked and checked and everything that’s in the queue,” Apfel explains. “It also shows how efficient warehouse workers are in picking and checking compared with their peers. We trace performance based on each worker and their badges. All inventory is moved through scanners, bar codes and badges, and the dashboard shows all that.”
Along with the above work, FragranceNet recently deployed an automated conveyor system from material handling systems provider Intelligrated that “eliminates a ton” of labor that was required for manual sorting, though he declined to be more specific. The retailer also wrote its own shipping-cost calculator and a system for grouping orders more efficiently—for example, by such characteristics as weight and delivery ZIP code—that has severely reduced the amount of walking required in the warehouse by workers there.
THE RESULTS: Besides the benefits listed above, the fulfillment systems overhaul has given FragranceNet the ability to ship up to 40,000 orders in a 24-hour period, a vital improvement for the holiday shopping season. “It’s pretty mature at this point, and it can handle Christmas,” he says. Apfel ticks off further improvements: “Orders are imported into our system, processed and picked and packed approximately 10 times as fast as our old system. We are able to prioritize orders in the warehouse. We are able to zone and batch pick which saves labor. We are able to automatically sort outbound boxes which saves labor. We are able to monitor all warehouse employee productivity, speed and accuracy throughout the workday.”
In fact, Apfel believes the in-house technology can support his plans for expansion. “The mantra we had when building it was: We want to be able to use it to sell any product,” Apfel says, adding that his company has a good shot of moving into new product categories. “We believe we can scale to $1 billion in sales one day,” he says.
Another benefit of this homegrown technology is that the retailer’s warehouse is almost completely paperless. That’s a “huge advantage, as there would always be a lot of labor in the print room.” Now that labor is free to do more important work, FragranceNet can keep its head count from swelling too much during the holiday season. Plus, were he or his employees able to identify a useful tweak, the work can commence immediately rather than waiting for an update from a vendor—one of the retailer’s initial motivations that drove its decision to build rather than buy.
THE COSTS: Apfel does not give an exact figure, but says the order management project alone ran “into the millions.” It’s difficult to pin down how much retailers can expect to spend on technology—factors such as size, product selection and site functionally can produce wide variations in pricing—but Gene Alvarez, an e-commerce technology analyst at Gartner Inc., says a fair rule of thumb for such spending is 5% of annual revenue. Forrester Research Inc., meanwhile, says that annual maintenance and upgrades of software typically cost five times the original price of building or buying the technology.
Of course, the costs of building in-house technology don’t stop with the creation of software or the deployment of new warehouse equipment. FragranceNet’s information technology department—the employees charged with building and maintaining the homegrown systems—has “dozens” of workers. “IT and shipping are our biggest expenses and investments, running at millions per year each,” he says without being more specific, though he does note that “under 2% of our revenue goes into IT spending.” That would translate into no more than $3.8 million in 2014, based on the retailer’s 2014 web sales.
THE CONTEXT: FragranceNet’s decision to build rather than buy—and the retailer reports having in-house e-commerce technology and systems for 18 separate e-retail functions, including comparison engine feed management, personalization and site search, out of the 30 measured by the Internet Retailer 2015 Top 500 Guide—bucks the broader e-retailer trend: More retailers are turning to vendor services and software rather than going it alone. For instance, 295 merchants in this year’s Top 500 report having homegrown order management systems, down nearly 9% from the previous year.
Several e-commerce experts, while not specifically familiar with the work done by FragranceNet, were generally surprised that a retailer would invest so much time and money into building its own technology. “The vast majority of retailers opt for either an enterprise or open-source [such as Magento] off-the-shelf solution for both e-commerce and order management,” says Peter Sheldon, a Forrester Research vice president and principal analyst. “Only a few brave retailers are building today,” he says. Among them, he adds, are such giants as Wal-Mart Stores Inc. and Staples Inc., which have “huge research-and-development budgets, huge development teams, an appetite for risk and a need to innovate faster than the vendor community can do for them. However, even these firms commonly use a combination of commercial technology with open-source frameworks and custom ‘build’ code. Their architectures are essentially a hybrid of build and buy.”
He offers that some smaller, well-funded and relatively niche e-retailers concentrate on building their own e-commerce technology—this might include flash-sale operators—“primarily because their culture and the CIO’s ego or philosophy calls for a ‘build’ approach. It can work for the first few years, but commonly those that set out on a build path inevitability run into problems with scalability and talent retention as they grow.” While he agrees that building technology in-house does give the retailer agility and flexibility, the costs are always “way more expensive in the long run,” and that it can get pricey to keep the talent in place that knows the custom system.
“We think we are extremely efficient when it comes to prioritizing and analyzing IT projects,” Apfel says. “We are always evaluating what the next build out should entail and what the ROI is for the new project or new features to an existing program.”
WHAT’S NEXT: Among the latest “bells and whistles” built by the FragranceNet IT department is an addition to the order management program designed to make its customer service operation run more smoothly: Enabling a customer service representative to see a customer’s history via a single tab on the browser to help the representative more quickly handle any calls or other concerns.
Despite others’ skepticism about the value of building versus buying, Apfel displays rock-solid confidence that his company has selected the correct path. “One of the reasons we survive,” he says, “is this build-versus-buy mentality.” The software FragranceNet has built, he adds, is “best-of-breed.”
For those retailers that want to imitate FragranceNet’s technology strategy, Apfel offers seemingly obvious but important advice. “Do a lot of research on the out-of-the-box solutions first,” he says. But, he adds, if you will need a lot of customized features, why not invest in your own technology rather than in someone else’s product?