Subscribe to our free, weekly email newsletter!


Warehouse and DC Best Practices: Sephora’s gorgeous network reorganization

Our warehouse/DC engineer dives into the best practices and tools that the retailer put to work in order to expand its distribution network after it became the exclusive provider of beauty products for JCPenney stores.
By Maida Napolitano, Contributing Editor
November 01, 2011

Drawing up the plan
Determining the best strategy required a systematic approach, the analysis of mountains of data using the latest database and network optimization software, the input of experienced team members with first-hand knowledge of the business and its future trends, along with some good, old-fashioned due diligence.

And it all didn’t happen overnight. The first two months were geared towards building a baseline model that mimicked Sephora’s current distribution network, followed by a few weeks of testing different logistics scenarios, culminating with a site-selection period of six months that winnowed the selection to the “perfect” site for the second DC. Here, the Sephora project team shares the steps to their success: 

1. Form an integrated team. First, a project team was assembled. It was led by Flaherty and his internal logistics team and worked closely with St. Onge’s network study team led by Bryan Jensen, vice president for the consulting firm. Input from key personnel from finance, transportation, operations, information systems, and marketing departments was then periodically required to provide the data, establish assumptions, and offer direction for future trends.

2. Understand business issues. The entire team had to understand and agree on how Sephora did business. How did replenishment to stores work? What did the stores need in terms of service? Was there a dollar value associated with having a same day service time to its clients? 

“Internal to the Sephora organization, we needed to make sure we got information, forward-looking expectations, and desires for the operation from the store managers,” explains Jensen. “That determines the boundaries in which the network analysis will examine how they might get their product to the marketplace.”

3. Develop baseline model “as-is network.” Over a four-week period, the team collected data from different areas of the business. Some of this data included a year’s worth of transactional history for its direct-to-consumer (Sephora.com) business and its store business, inbound and outbound freight costs, warehouse operating costs, shipment volumes, and store locations. For most of the preliminary data analysis, St. Onge used SQL query tools that allow the users to manipulate massive data files. 

St. Onge then spent the next four weeks building a baseline model of the existing network. It used leading-edge software specific for distribution network optimization to bring all the above data and assumptions together. The goal was to validate the baseline model by replicating the current distribution network, applying the appropriate transportation costs and volumes, and comparing it against last year’s actual historical costs. The costs determined by the model had to come within a very close one percent. 

4. Establish logistics objectives. After the baseline model was validated, the team began to establish the objectives for the future network. According to Jensen, trying to nail down predictions for Sephora’s future was the most challenging step. The team needed to understand not just the percentage growth in new stores per year, but where they were going to open these new stores as well as the plans for growth of their direct-to-consumer operation.

Would it geographically follow the store patterns and the store population throughout the country? Typically the standard is to look five years ahead, but Sephora also provided St. Onge with a 10-year outlook. “Clearly the further out you look the fuzzier your visions gets,” says Jensen, “but at least you can understand directionally how things will trend beyond a normal five-year horizon. This can be important in a network analysis because implementing the solution can take considerable time.”

5. Identify logistics modeling scenarios. Once growth projections and other future logistics requirements were entered into the model, the team then identified two main scenarios that they wanted to test. Scenario 1: What’s the best East Coast site if Belcamp is closed? Scenario 2: If Belcamp is fixed, what’s our best second site?

For each scenario, the model was populated with statistical data regarding candidate locations. This data includes the average cost per square feet of a DC at that location and all the freight rates to and from that location. The model then rates the candidate sites and ranks them based on the cost to service. 

6. Model scenarios and evaluate. With Scenario 1, the model was run with a clean slate to determine whether or not Belcamp was indeed the optimal site. It turned out that the absolute optimal location was just outside of Philadelphia on the New Jersey side of the Delaware River near Cherry Hill, which is only about 80 miles from Belcamp. “When you’re that close you don’t bother relocating for the amount of transportation costs that would actually be saved,” says Jensen. Management then decided to extend Belcamp’s lease.

In Scenario 2, with Belcamp fixed, the team then re-ran the model with an eye towards optimizing transportation, lead time, and expenses with a second site. “It put us in the general area of Nevada, New Mexico, Colorado, Arizona, and Utah,” says Sephora’s Flaherty. “From there, we reviewed a variety of secondary criteria: the demographics; the cost of doing business such as business licenses, permits, tax credits, incentives from the state; and utility costs.” In the end, Salt Lake and Reno were neck and neck.

7. Prepare an implementation plan. Over six months, Flaherty began the task of implementing this two-DC network solution, personally travelling to both areas, checking out different buildings, and weighing out strengths and weaknesses of each site.

“While both cities looked very promising, Salt Lake City not only optimized our transportation costs, but the local and state governments were very responsive and eager to work closely with us to ensure that our facility was brought online in the shortest amount of time possible,” says Flaherty. He adds how the Salt Lake City’s entire business community never wavered in its support to bring Sephora to the city along with the jobs that it offered the area residents. 

A network with benefits
It’s been three years and Sephora’s two-DC network has significantly improved its customer service cycle time. “Because I’m closer to my stores,” says Flaherty, “I could get replenishment faster, making it less likely to go out of stock on a particular item.” It has also achieved freight economies, while relieving capacity at the Belcamp DC.

Jensen points out another “priceless” advantage with the opening of this second DC. “When you have only one building, it’s a critical point of failure if a fire or a flood devastates it,” says Jensen. “Having two facilities engenders a level of business continuity or additional redundancy to the network.” 

And Sephora continues to grow. “Last year we purchased the largest beauty retailer in Brazil, and we have plans to expand Sephora into that country beginning in 2012,” says Flaherty. “Towards the end of this year, we will be expanding into Mexico.” 

Ever vigilant, the logistics team has just completed another network study to determine the need for a third facility—and so the cycle continues.

About the Author

image
Maida Napolitano
Contributing Editor

Maida Napolitano has worked as a Senior Engineer for various consulting companies specializing in supply chain, logistics, and physical distribution since 1990. She’s is the principal author for the following publications: Using Modeling to Solve Warehousing Problems (WERC); Making the Move to Cross Docking (WERC); The Time, Space & Cost Guide to Better Warehouse Design (Distribution Group); and Pick This! A Compendium of Piece-Pick Process Alternatives (WERC). She has worked for clients in the food, health care, retail, chemical, manufacturing and cosmetics industries, primarily in the field of facility layout and planning, simulation, ergonomics, and statistic analysis. She holds BS and MS degrees in Industrial Engineering from the University of the Philippines and the New Jersey Institute of Technology, respectively. She can be reached at .(JavaScript must be enabled to view this email address).


Subscribe to Logistics Management magazine

Subscribe today. It's FREE!
Get timely insider information that you can use to better manage your
entire logistics operation.
Start your FREE subscription today!

Recent Entries

It’s small, but senior brass at YRC Worldwide will take it. After nearly seven years of continuing losses in excess of $2.6 billion, the parent of the nation’s second-largest LTL carrier posted a narrow net profit in the third quarter ended Sept. 30.

As was the case for the second quarter, third quarter earnings results for publicly-traded less-than-truckload (LTL) carriers are again strong. Signs of solid earnings results from carriers that have posted earnings to date include tonnage increases, gains in weight per shipment and average daily shipments, higher yield, and revenue per hundredweight.

While the holiday season is known to bring good tidings and cheer to all, it may also come with another thing that is not so pleasant: higher rate freights. That was the thesis of a commentary written by Mark Montague, industry pricing analyst and chief market-watcher for DAT, a Portland, Ore.-based subsidiary of TransCore.

Earlier this week, FedEx said it is expanding its International First service for early deliveries with the addition of 31 new origin countries, which will bring the total number of origin markets for the service to 97.

Monday, December 22 is pegged as UPS's peak delivery day, as the company expects to deliver more than 34 million packages that day, adding that it expects to see six days in December top last year’s peak shipment day delivery record of 31 million packages.

Comments

Post a comment
Commenting is not available in this channel entry.