During the 1999 Christmas shopping season, consumers placed an estimated 36 million online shopping orders—more than four times the orders recorded during the same period the year before, according to BizRate.com. The Wall Street Journal reported that online sales totaled an estimated $3.35 billion between Nov. 26 and Dec. 16, 1999.

When those online holiday shoppers make the transition to ordering everyday items like groceries, pharmacy products, and household necessities online, the online grocery business will explode. The potential market opportunity for online purchases of food and household goods and services is enormous. By 2007, Andersen Consulting projects that 15 million to 20 million U.S. households will use such services. That's more than 15 to 20 times the number of households using online grocery services today.

Given this expected growth, Peapod, the United States' leading online grocery service, must continually re-examine every aspect of its business. The marketing, pricing, distribution models, customer incentives, and hiring practices that worked for a Chicago-area start-up back in 1991 now need to support a business with customers in eight major metropolitan markets (Chicago; San Francisco; Boston; Columbus, Ohio; Long Island, New York; and Dallas; Houston, and Austin, Texas). To keep up with our customers and their ever-rising expectations of e-commerce, we are continually evaluating and changing every aspect of the online grocery business.

This article will describe our efforts to focus our supply chain more strongly on fulfilling customer requirements. In addition, it will discuss some of the differences in logistics management for consumer-direct e-commerce, compared with more traditional industries where I worked for many years in logistics and supply chain functions.

So far, there seem to be as many different ways to manage the supply chain for online grocers as there are online grocers. At Peapod, we've designed our supply chain to maximize flexibility for the customer. That means that our customers have more choices built into our supply chain as a result of these strategic decisions. For example, a traditional logistics operation might schedule weekly deliveries of bulk quantities—cases and pallets—to a store in your neighborhood. With Peapod, the consumer chooses the day and time of the delivery. Peapod adapts to the consumer's needs; not the other way around.

As the following list shows, e-consumers are demanding. Here's what our customers want:

• Choice of delivery times and formats.
• Choice of product mix.
• Low cost.
• A complete order.
• Timely delivery.
• Fresh and damage-free produce, meat, and other products.
• Accurate and complete invoices and packing slips.
• Up-to-the-minute information on order status and problem resolution.

Expectations of customers—whether businesses or consumers—get more exacting every day. As this list shows, these expectations are particularly evident when dealing with the direct consumer.

With all of these requirements, the variables are phenomenal. Because logistics is all about trade-offs, Peapod needs to look at all those trade-offs and balance them. And as the market and customers' expectations change, Peapod needs to adjust its logistics practices continually.

For supply chain managers—particularly those who have worked in the profession for a while—the challenges associated with direct-consumer delivery require a different way of thinking. Compared with traditional industries, the logistics demands of consumer-direct e-commerce are fundamentally different in three important ways: quantities, timing, and demand management.

• First, single items are replacing bulk shipments. In the past, supply chain managers always concerned themselves with moving cases and pallets of product. You received everything in cases and pallets. You took orders in terms of cases and pallets. But now, with dot-com companies, everything is an each-pick—that is, picking single items. You're choosing and shipping one loaf of bread, one quart of milk, one vial of dental floss, and so on. But, this newly granular supply chain actually builds on traditional principles of logistics management, and the warehouse employee's job is fundamentally the same. Before, he or she was told to go to a location and scan a carton. Now, the employee goes to a location and scans a loaf of bread. The basic equipment, such as carousel racks, flow racks, and conveyers, is the same. The record keeping is in terms of individual units but, otherwise, is not much different. Because items are not protected by racks and boxes, more care must be paid to handling the items and assuring that equipment is appropriate for the particular items.
• Second, in the past it has been acceptable to get shipments to customers within 24 to 72 hours. For many dot-com companies, that's not good enough. Particularly for online grocery services, consumers want immediate delivery. So Peapod is constantly striving to reduce the time from order to delivery. In addition, Peapod enables customers to choose their delivery time slot, so our delivery window is that much smaller. We aren't aiming to deliver your groceries sometime on Tuesday; we have promised to deliver your groceries within a specific time window on Tuesday. These time commitments eliminate some of the routing flexibility present in a traditional logistics environment, and they place added pressure on drivers to meet intraday deadlines.
• Third, it can be very difficult to match customer demands to corporate resources. Many entrepreneurs have found that if they put up a Web site, they get pounded with orders—sometimes, more orders than they can fulfill. The need to match demand to capabilities means that marketing and logistics must work hand in glove.

These differences mean that players in this industry don't have as much flexibility in logistics planning, compared with a traditional company. Logistics decisions are driven by our overall strategy and by market and customer requirements, rather than solely by cost comparisons for different routes or truckload planning, for example. One important principle of supply change management still holds: Putting in place and training the right management team and employees is a critical element of success.

Because customers have more choices, logistics options are somewhat more constrained than those of a traditional company. Therefore, we find ourselves working more closely with marketing to influence customer behavior and, thus, improve supply chain efficiencies. Of course, logistics and marketing functions should work closely everywhere. But this doesn't happen all of the time in all companies. At Peapod, we're joined at the hip. I spend more time with marketing than with any other function. That's because marketing is central to solving our key logistics challenge: making deliveries efficient.

It is a constant balancing act to achieve the optimum efficiencies in deliveries because customers drive Peapod's delivery schedules. Customers choose the day they want their groceries. Customers choose a delivery time slot. If customers place an order by 4 p.m. on Monday, they can arrange for a delivery between 7 a.m. and 1 p.m. Tuesday. If they order by midnight Monday, they can have their groceries between 4 p.m. and 10 p.m. Tuesday. Of course, perishable food poses additional logistics challenges.

This flexibility and consumer choice is very different from the ironclad schedule of the milkman of old. The milkman stopped at every house, every day, or on predetermined days. The milkman's route never varied. He was a take-it-or-leave-it vendor. In contrast, Peapod aims to maximize the consumer's flexibility.

To get the most flexibility for consumers in the most cost-effective way, logistics and marketing have no choice but to collaborate. For example, once we have a customer in your neighborhood, we'll groom and cultivate the neighborhood through marketing to get other customers nearby and make the delivery route more dense.

Ultimately, we may offer myriad delivery and pickup options to suit different preferences; for example, drop-offs in a corporation's parking lot, drop-offs at a central pickup point like a retail store, and unattended drop-offs at homes. Options like these will make it possible for drivers to drop off more orders in a given shift. Increasing the density of stops increases capacity, which makes Peapod more like the milkman and, thus, more cost effective.

The inherent inefficiencies of each-picks and the constraints on routing make the online grocery business a real logistics challenge. We're attacking this challenge aggressively by thinking outside the box. This industry is like Wilbur and Orville Wright. Who knows what that 747 will look like? Today, all of the e-commerce companies are pioneers. The methods we're using today could be very different from the supply chain models used five years from now.

And in fact, Peapod's distribution model today is quite different from its original distribution methods. Our original distribution model took the consumer out of the grocery store but did not eliminate all of the inefficiencies of the grocery store. Instead of the consumer, a Peapod "shopper" walked through the store aisles selecting staples and perishables. Over the last 10 years, Peapod has experienced substantial order growth. Through 1998, our goal was to have Peapod shoppers essentially mirror consumers' behavior in supermarkets. Using Peapod's picking system, they shopped individual customer orders in separate areas of supermarkets. A replenishment manager employed by the supermarket was responsible for stocking; Peapod employees did the picking. In the early years of our company, this shopping model had many advantages: low fixed costs, flexibility to expand, no inventory, and no facilities-management issues.

A distribution model that made sense in 1990, when Peapod began, became somewhat outdated by the late 1990s with volume skyrocketing. Peapod needed to revise its approach to inventory, warehousing, order fulfillment, route planning, and scheduling. As volume and competition increased, the shopping model needed an update.

For Peapod, it became a case of back to the future. By looking at our business as a fulfillment and distribution business, instead of as a shopping business, Peapod has not only reduced the cost to fulfill an order but also turned around orders more quickly and shipped more complete orders—a prime strategic requirement.

This new distribution model was developed in response to our customers' demands. Peapod's customer, typically a professional woman with children living at home, places an online order via our Web site, www.peapod.com. She orders a mix of dry, refrigerated, and frozen products. The average order has 40 to 50 line items and an average revenue per order of $120. Orders placed the day before are delivered the next morning, and orders placed the night before are delivered the next evening. Customers also can request delivery at any time within the next week. Peapod delivers groceries seven days a week.

We're working aggressively to reduce the time between order cutoff and delivery. In e-commerce, standards of customer service are improving constantly. Some grocery e-tailer is going to shorten the window for pick and pack; Peapod wants to be the one. We plan to offer same-day delivery at some point in the future.

Orders pulled off the Web site are put immediately into our routing system. This best-of-breed customized software incorporates maps and travel time to determine how many stops each vehicle can make the next day. We use sophisticated Web-site technology that applies airline reservation principles to consumer-direct delivery and orchestrates supply and demand to minimize transportation costs. Furthermore, we accomplish this while fulfilling the service promise and leveling orders across days and shifts. These efforts make our load planning and routing second to none. As orders come in, we can keep adding to different routes. We'll pass the time savings of this improvement on to the consumer by extending the cutoff times, thus giving the consumer more time to order. That's our goal: We want our process improvements to improve the customer's experience, not merely to cut our internal costs.

The route-planning software creates a route and stop number and provides maps, point-to-point directions, and time standards for the driver. Typically, each distribution center will handle large order volumes for each of two daily delivery periods (morning and evening).

Consumers are fussy about their groceries. They want a certain amount, a certain degree of ripeness, a certain size, freshness, broad selection, and so on. To succeed as a grocery e-tailer, Peapod must provide the same quality of product that consumers would choose themselves. In a sense, Peapod shoppers become our customers' alter egos.

But all of these factors are operational benefits to Peapod, not necessarily added value for customers. As our volumes increased, it became apparent that we needed a distribution model that focused more on our customers' main concerns: product availability and quality. We needed more control over both.

However, the retailer was having a hard time replenishing the shelves. This has been our customers' prime complaint—that is, not receiving a product they have ordered. Because Peapod is a time-saving service, not merely a shopping service, out-of-stocks are completely unacceptable. For example, say a customer orders all of the ingredients for a chili dinner: ground beef, rice, peppers, spices, garlic, onions, tomatoes, tomato sauce, and red beans. If we deliver all of the items but the red beans, which are out of stock, the consumer still has to run out to the store. As this example shows, one out-of-stock item can defeat the whole purpose of the value proposition to consumers.

To minimize out-of-stocks, we're reorganizing our distribution model. We now use a central distribution center in two of our eight markets (Chicago and San Francisco). In all new market areas, we are using stand-alone distribution centers with separate inventory and replenishment systems. This way, Peapod selectors don't compete with supermarket shoppers for stock. By handling our own ordering, receiving, stocking, and replenishment, we can fulfill customers' orders more completely. We also have the capability of ordering directly from wholesalers. Because of the quick turnaround in the grocery industry, we essentially have a just-in-time inventory in the distribution centers. That is, based on orders by 4 p.m. Monday, I can place my perishables order for Tuesday. We use a movement report to identify products that aren't moving and suggest to marketing that these products be dropped.

As we build new warehouses or retrofit existing facilities, we plan to use a common layout. Our preferred size is about 100,000 to 125,000 square feet. Some of our competitors are building facilities double and triple our square footage, but we're following a more conservative investment strategy. Generally, we're striving for a balance between automation and people. We want to avoid processing problems that could stem from automating too much too early.

Our "selectors" pick and pack the morning orders overnight, finishing at 4 or 5 a.m. Another shift packs the evening deliveries beginning at 7 a.m. Selectors use a specially designed cart with an insulated tote. The selector simultaneously chooses items and packs the bags. Using our paperless picking system, the orders are received by the selectors electronically, with labels indicating the tote with route and stop number. Then, they pick with scanners, just as if they were picking pallets or cases in a traditional warehouse. If the scanned item doesn't match, selectors cannot go on to the next item.

Essentially, we use a combination of zone picking and batch picking. Some people pick perishables; others pick dry goods. Dry, refrigerated, and liquor orders are picked separately. For example, a selector might pick all of the cereal for the next 10 orders. Perishables pickers are experts who are trained to examine the produce. The Peapod selector is the consumer's eyes, nose, and hands. We make sure the perishables are fresh and of the finest quality.

Throughout our business, choosing the right people is a key element of success. Previously, when Peapod needed supermarket workers, we advertised for a "shopper." But the concept of a shopper doesn't really describe the sort of culture and attitude we're looking for in a distribution center. Picking groceries may not require a forklift, but it is hard work. We need a person who can work a night shift, who can follow instructions and scan, and who can learn about the products.

Because we're making each-picks, we're continually adjusting our processes and procedures to develop the most efficient ways to pick. In our distribution centers, we want the aisles tight, so people don't have to walk a lot. We devote a lot of time to design and layout of the distribution center and to materials-handling equipment and systems, just as a traditional distribution center would. Because we're picking from a distribution center, not a store, the selectors work in an environment designed to maximize their performance—not an environment necessarily designed for merchandising to consumers. So, for example, we'll put the toothbrushes and toothpaste together because consumers are likely to order them together, as in a grocery store. But we also organize products into A-B-C categories, based on frequency of ordering. In addition, we arrange products by type, not by brand, which is different from what you would see in a grocery store and more akin to traditional distribution center procedures.

My father was in the supermarket business for 33 years; I started as a bagger at the age of 13. He taught me that you should never see a pallet or any cases during shopping hours. We have the same challenge in our distribution centers. We don't want people replenishing while the selectors are picking and packing because we need to make the best use of the selectors' time.

The dry, refrigerated, and liquor orders are combined in a large staging area organized by route, with the floor permanently marked for every route. Loaders match route and stop numbers on the totes in the staging area. Totes are scanned as they are loaded in the trucks. As we add more routes and vehicles per distribution center, we will need a large, well-organized space for staging. Cross docks will be used as a strategy to be closer to our customers, as well as to minimize costs.

As we rent or build stand-alone distribution centers, our goal is to make the design of these spaces and of the facilities for maintenance and vehicle management consistent from one distribution center to another.

The truck driver works from a route summary sheet that includes destinations, directions, and arrival and departure times for each delivery. Drivers are in constant contact with the dispatcher to be sure we meet customers' time slots.

In a traditional logistics operation, a driver is a driver is a driver. He or she drives a truck from point A to point B according to a prearranged schedule.

At Peapod, a driver is the only face-to-face link to our customers. The driver is absolutely crucial to the way customers perceive our company, to their overall satisfaction, and to our understanding of the customers' needs and preferences. That's why we call drivers "ambassadors": They represent Peapod to customers as no one else can.

Ambassadors not only drive. They listen to customers' comments. They accept coupons. They follow special delivery instructions. The driver is an accountant, a driver, a marketer. This expanded role requires a different set of skills and experiences, different job descriptions, different training, different incentives, and different forms of feedback and evaluation. Ambassadors must pass background tests, drug tests, and driving tests. In addition to incentives for safe driving, punctuality, and attendance, there are incentives for repeat business and on-time deliveries (if within their control). We keep schedules consistent and teach ambassadors how to present groceries at the door. Typically, a team of two to three drivers will cover the two-shifts-per-day, seven-days-per-week schedule for a given neighborhood. (Exhibit 1 summarizes the order-fulfillment process from keyboard to front door.)

I've been working on logistics for Peapod for about nine months, first as a consultant and now as chief logistics and operations officer. In that time, I have found that, although the details of delivery differ, the overarching best practices of logistics management apply equally in the dot-com environment. In fact, many of the metrics we use at Peapod are common to traditional companies. For example, we measure productivity by shift and selector, cost per order, and output (for Peapod, an "each"; for a traditional company, a case or pallet).

Most of the changes we are implementing and have implemented reflect best-practice principles of supply chain management for any type of company. The following six principles are among the most important.

1. Every logistics function—dot-com or traditional—needs the right management team. Especially at the level of the distribution center, management is key. That distribution center is a small business, and it has all of the functions of a business: human resources, finance—everything you can think of. Therefore, at Peapod, we screen employees carefully and do background checks when hiring. Facilities managers get at least three or four weeks of training over the course of a year on topics including day-to-day job duties, supervisory skills, and important human-resources issues like sexual harassment and drug screening. Each facility has its own human-resources manager, who is part of the facility's management team and helps with recruiting.
2. People are the most important element of every company and every logistics operation. We've all heard many times that when people understand clearly what is expected, they will work hard to meet goals. This advice is common because it is true. At Peapod, we have adopted written job descriptions that specifically describe what a selector does, what a loader does, what an ambassador does. These job descriptions help us not only to supervise current employees but also to hire new employees who are well suited to the requirements of each job. We invest heavily in employee training for every position, from facility manager to ambassador to customer-service representative. We have full-time trainers, and the two stand-alone distribution centers have trainers on site all of the time so they can follow up if something doesn't seem right on the floor.
3. Personal accountability assures high quality. Peapod wants its customers to receive the best-quality products and services. By making a produce supervisor, for example, responsible for quality, we assure better-quality produce. By posting expectations for productivity, currently by employee number but in the future by employee name, we let people see how their work compares to others'. By rewarding both teams and individuals, we keep individual commitment strong.
4. Statistical principles apply to each-picks, just as they apply to traditional warehouse management. It is important to measure and evaluate every step of the distribution process, even though, at Peapod, these steps are different for every customer and involve single items. We have been changing the layouts of our distribution centers by grouping the most popular items, applying Pareto's 80/20 law. And although Peapod selectors are walking aisles, not operating a forklift or driving from Akron to Cleveland, we still must minimize their travel time and minimize the number of touches to a product. The less we touch a product, the better off we are, in terms of productivity, selection errors, and damage.
5. Cost consciousness must be built into every strategic decision. Consumers like the convenience of online grocery shopping but they may not necessarily like the price. That means that we have to pursue cost savings aggressively. Increasing volume and creating denser routes, for example, are essential to lowering costs. In addition, it's important to encourage all associates to think outside the box and to make suggestions for improvement.
6. Focus on the customer. Last, but most important, the purpose of our business is to create and keep a customer. To do that, we must do the things that will make people want to do business with us. We need to provide value-added services that will reduce the consumer's overall cost, increase the consumer's productivity, and make the consumer's job of buying groceries easier.

Groceries are the cornerstone of Peapod's business, but they are just the beginning. Some day in the future, Peapod may offer one-stop or no-stop shopping for groceries, pharmacy products, pet supplies, toys, stationery, and so on. We aim to be the company that delivers to the household between two and three times per month. If we decide to handle these other products, it would be through partnerships and using cross-docking arrangements.

At Peapod, we want our customer to buy and buy again. So we have to make the customer's experience wonderful in every way. Our customers buy not only product but also service. Because service and product go hand in hand, continual improvements in warehousing and distribution are a key element of our business strategy.

For logistics professionals like me, this business represents a great opportunity to change today's supply chain paradigm. It's a chance to re-energize and apply principles we have learned in other companies in a new, electronically enabled environment.