Executive Briefings

Amazon.com Sees Supply-chain As Crucial To Its Future

Key Amazon.com executives in Seattle and Europe - including founder Jeff Bezos - talk about how the company's supply-chain strategies and fledgling European operations are shaping its prospects.

The next six months or so are going to be the most critical in Amazon.com's short but spectacular history. By then, reckon such top Wall Street analysts as Merrill Lynch's "New Economy" guru Henry Blodget, it will become clear if Amazon.com's business model will generate sustainable profitability and positive cash flow.

The implications are far reaching. Amazon has become something of a bellwether of the New Economy as a whole. If Amazon makes it, expect the rest of the dotcom crowd to mirror its strategies. If it doesn't - well, the bloodbath will make NASDAQ's April and May decline of 35 percent or more seem like a picnic. Amazon's supply-chain strategies and its fledgling overseas expansion could well determine the outcome.

The facts are simply stated. In a little over five years, Amazon.com has captured 20 million customers in 150 countries - 10 million of them in 1999 alone. To sleepy physical world retailers - you know, those sticks-in-the-mud who still have only shops - "Amazon" has become a verb, and one that increasingly generates the kind of fear associated with the news that Wal-Mart is going to turn up on your doorstep.

For founder and CEO Jeff Bezos, it's sweet vindication. During the five years that Amazon has been rewriting the rules of retailing, he recalls, the company variously has been called Amazon.con, Amazon.toast "and my own personal favorite: Amazon.org, because we were clearly a not-for-profit company." Not bad for a quietly studious investment banker who'd headed to work each day at a desk in New York, until, in 1994, he'd spotted some statistics on the growth of something called the internet.

Yet that success has a darker side: Amazon's famous unprofitability. For all its success compared with other recent dotcom fiascoes, the fact remains that its spiraling revenue, which reached $1.6bn last year, has been matched by equally spiraling losses: $31m in 1997, $125m in 1998, and a massive $720m in 1999 - almost six times the previous year's figure. And second-quarter 2000 losses reached $317m, prompting yet another collapse in the stock price as investors fretted over slowing sales growth and rising distribution costs.

Bezos remains sanguine - if visibly more irritated with the doomsayers who write the company off. "E-commerce is a scale business, with costs that are largely fixed," he insists. And the classic solution to dealing with a high fixed-cost base? Volume - and lots of it.

So Amazon is expanding fast from its core books, music and videos beachhead. Consumers clicking onto Amazon.com these days find themselves facing an almost bewildering variety of offerings: electronics, software, toys and video games, lawn and patio products, health and beauty items, kitchenware, tools and hardware, pet care, home furnishings - some 18 million unique items with more being added all the time.

"Two years ago, we were U.S.-only, and books-only," Bezos observes. "We weren't selling music, or videos, or any of the other product categories we now sell. Two years later, although our U.S. books business has continued to grow rapidly, more than half of our revenue comes from these new categories and businesses." And with every additional dollar, the fixed costs of the distribution centers and headquarters systems are amortized that little bit further.

And make no mistake: The fixed cost base that Amazon has incurred with its distribution operation is a hefty one. In the last year, explains Jeffrey Wilke, vice president and general manager of operations, and a veteran of AlliedSignal and Andersen Consulting, the company has increased its worldwide distribution capacity from 300,000 square feet to more than 5 million square feet. The combination of rapid customer growth and rapid product line expansion, he explains, meant that Amazon's original two distribution centers - one in Seattle, one in Delaware - simply couldn't cope. The original 93,000-square-foot, largely manual facility now seems ludicrously tiny for a business of Amazon's size. Even the 202,000-square-foot facility in New Castle, Del., which doubtless appeared a brave gamble back in November 1997, now seems woefully undersized.

So four further U.S. distribution centers have opened in the last year: one in Fernley, Nev.; one in Coffeyville, Kan.; one in McDonough, Ga., and one in Campbellsville, Ky. (A further facility came with the acquisition of e-retailer Tool Crib last October, which at the time of writing is still being used solely to handle tool orders.)

Highway Automated DCs
Each of these distribution centers, says Wilke, has roughly 700,000 square feet of warehouse space, although that's an average that varies widely, as does each distribution center's provenance. Stanley Tools formerly owned the 320,000-square-foot Fernley facility, located in the Nevada Pacific Industrial Park some 30 miles east of Reno along Highway 80. On the other hand, the greenfield Coffeyville distribution center in Kansas, at 850,000 square feet is more than two and a half times the Fernley facility's size.

The distribution centers are highly automated, using barcodes and scanners to track and sort inventory that moves through the facilities. A typical Amazon DC has 10 miles of conveyor belts and ships most in-stock items within 24 hours of the order click.

But for there to be a tomorrow for Amazon - and remember the battering the company's stock price has taken in recent months, slumping from around $115 to a touch under $35 - the company has to suck in still more volume to amortize its high fixed costs. With growth in the core U.S. books and music market slowing down, and new product areas still at the fledgling stage, the sources of this extra volume are limited.

Which is why the company's management has started to expand the business geographically, as well as into new product areas. Amazon now has a U.K. subsidiary, headquartered in Slough, west of London, employing around 500 people - Amazon.co.uk - as well as a slightly smaller German one, Amazon.de, headquartered in Regensburg, Germany.

More are planned. Bezos is tight-lipped about where precisely these will be, but executives such as Wilke drop broad hints: Look at national levels of internet penetration, and overlay per capita disposable income figures, he suggests. Certainly, the establishment of a Europe-wide call center operation in The Hague suggests that further European ventures are close to completion.

Identifying desirable global locations for new distribution centers is one use Amazon will make of new supply-chain software from Manugistics of Rockville, Md. Amazon announced in May that it would install Manugistics' NetWORKS solutions to support its global expansion and operational improvement initiatives. It will use NetWORKS Strategy to model fixed and variable network costs, taking into consideration such factors as varying transportation and supplier lead times, and global constraints such as tariffs and taxes. The model will then be used to design an optimal global network.

Amazon.com's initial steps toward international expansion appear to be going well. Market research numbers from Media Metrix place Amazon.de and Amazon.co.uk as the most visited e-commerce site in their respective countries, and in the top 10 of sites visited overall in each country. But neither operation - either individually or together - will make or break Amazon on its own.

Instead, they serve as a stark test-bed. The dilemma is simply stated. To survive, Amazon must increase its volume by succeeding in performing a trick that most U.S. retailers find challenging: translating a highly successful U.S. format into one that works just as successfully overseas. And most don't. In recent years, for example, Toys 'R' Us, Wal-Mart, Staples and others have experienced disappointing results when transplanting their U.S. business model to foreign soil.

Is Bezos worried? Amazon's core offerings, he is convinced, will play as well in Peterborough as they do in Peoria. "When you move to a new country, you never find that people say, 'Could you make this [web site] a little harder to use?' It doesn't happen that way. Nor do they say, 'Can you offer less choice, less selection, and higher prices?' The core things really are the same," he asserts.

It is, of course, too soon to tell. But the signs so far are promising, according to Steven Frazier, Amazon.co.uk's managing director, despite the fact that losses as a percentage of sales are declining slowly. In The second quarter, the U.K. and German operations together lost $35m on sales of $73m, compared with $16m on sales of $31m for the same quarter last year.

Overall, the German and U.K. operations made $167m in sales in 1999, Frazier points out, up from $22m in 1998, and are currently achieving a run rate of $280m a year. Both businesses were acquired in 1998 from existing online book retailing companies - in the U.K.'s case, an outfit called BookPages - and Amazon has been careful to retain key staff and local market knowledge, he says. Despite the similarities between the Amazon.com and Amazon.co.uk web sites, "we're trying to be a local business serving a local market."

Nevertheless, there are differences. Forget the one-stop trip to the virtual mall, for example. Amazon.co.uk merely sells books, videos and music - the core Amazon.com formula. Still, for U.K. and German customers deprived of large U.S.-style physical bookstores (the vast majority of the population) Amazon's range is tantalizing: Amazon.co.uk sells 1.2 million British books in print, more than 250,000 U.S. titles, 220,000 CDs and 23,000 videos. Other products will come, but will take time to do so, Frazier concedes. "The limits to what we can do are huge, but on a day-to-day basis there's a brutal process of prioritization: What do you do next?"

For one thing, Amazon.co.uk began offering software and toys, driving sales by awarding existing customers free gift vouchers on top of hefty discounts.

Growth already has driven the U.K. operation from a small distribution facility close to the Slough HQ to a 220,000-square-foot distribution center in the Marston Gate logistics park in Milton Keynes, a hub north of London and on the country's main arterial freeway, the M1. Both operations were kept going to cope with the busy Christmas period, but then Milton Keynes took sole charge. And even this is a temporary option according to Stuart Armstrong, the company's U.K. head of operations, while a more permanent home, with a level of automation comparable to that found in the U.S., is being erected.

Are there any differences between the U.K. and the U.S. modes of operation? Although the basic philosophy - high levels of automation, for example - is the same, Europe's relatively compact nature, at least compared with the U.S., also has proved a boon. In contrast to the U.S., where the distribution model revolves around large distribution centers located equidistant from major population centers, Europe allows Amazon to play more sophisticated tunes.

Take its proximity to suppliers, for example. The Marston Gate distribution center is close to the warehouses of several book, music and video suppliers, most of who deliver twice a day, Armstrong points out. If an item is out of stock, when a customer orders it, it can be shuttled over on the next delivery, and still sent on time to the customer. And whereas 32 percent of Amazon's U.S. shipments are funneled to parcel carriers like FedEx and UPS, that is not necessary in the U.K., where the Royal Mail's standard service is 1-day delivery. "No country ever appreciates its local postal system, but the Royal Mail and Deutsche Post are among the best in the world," observes a happy Bezos, genuinely delighted to find the opportunity to offer customers an even better deal.

And in some cases, a deal that is not yet possible in the U.S. - same-day delivery, at least within London. "We began offering it through a private carrier, but then found that Royal Mail would offer the same thing," says Armstrong. So, provided that customers order within a given time window, they are offered the option of same day delivery as a free upgrade.

Positive Cash Flow
Will it, in the end, be enough? Even with U.S.-style 50 percent discounts and the rest of the Amazon customer experience? Certainly, with the share price as it is, customers are just about the only people who are feeling bullish at the moment. Employees, managers and investors alike are bruised and battered. Suppliers, too, will be heeding analysts' fears over running out of cash - fears that Amazon dismisses with flat denials and projections of a positive cash flow by quarter four.

One of the targets of Amazon's critics is its investment in distribution centers. Why incur the expense of establishing such operations at all, they ask. In a business struggling toward operating profitability, why add fixed leasing and employment costs to the cost structure when the use of third-party contractors would permit a much greater degree of variability - not to mention allow Amazon to focus more on its core specialty, online retailing?

Quite apart from the delicious irony that physical distribution is Amazon's only profit center on a straight cost basis (the 1999 accounts reveal that it charged its customers $239m for shipping their goods to them, but only incurred shipping costs of $227m), there is a strategic issue at work.

At its heart, Bezos explains, is the belief that "the end-to-end, 'click-to-ship' customer experience that Amazon creates is even more important in the online world than it is in the physical world." Why? "Online, the word-of-mouth impact is amplified," he explains. "Every internet customer has a big megaphone, and if we make a customer unhappy, they don't tell five friends, they tell 5,000. And the reverse is true: You create evangelists. If you do a great job for customers, then they tell 5,000 friends, in newsgroups and listservers and chat areas, etc. - so that online, your marketing dollars are best spent building great customer experiences."

And contrary to more recent arrivals' massive marketing and advertising budgets, Amazon's promotional expenditure does indeed remain relatively low for a business of its size - $141m in 1999, up from $60m the year before. "In our first year, we did not do a dollar of paid advertising, and we still grew incredibly rapidly," Bezos says, pointing out that in its first six months of trading, Amazon had shipped books to more than 80 countries. After a year, the number of countries shipped to had climbed to 150.

"Today," he adds, "we do a significant amount of paid advertising, but it still accounts for less than half of our new customers." Time after time, he says, when probing the reasons why new customers came to Amazon, company studies have shown that more than half the new business comes from word-of-mouth referrals.

So, goes the logic, every element of control that Amazon cedes over its operations is an element where that customer experience can be put at risk. Wilke puts it this way: "We think that it is because we control the entire process from click-to-ship that our customer experience is what it is." While the company has of course looked at third-party operators, he says, it has always come to the view that "what we do ourselves we'll do better than the market can offer right now."

In part, the issue is the creation of a sense of ownership, exemplified by Bezos's insistence that every employee in the transaction chain, however lowly, is compensated in part with stock options - and the greater a person's impact on the overall customer experience, the higher the proportion of stock options that constitutes their salary. "If you want people to behave like owners, you have to actually make them owners. There are no shortcuts to that process," he says.

For proof, notes Wilke, look no further than last Christmas - the first real e-Christmas, according to many pundits. It was, for many B2C e-commerce companies, an enormous public relations disaster, with a huge number of column inches devoted to how angry consumers found themselves without the products that they'd ordered. "If you look at those sites," says Wilke, "many of the ones who had trouble were using third-party logistics contractors. We decided to do it ourselves - and spent a great deal of money doing so - but by December 24th every distribution center was completely clean: we'd shipped 99 percent of our orders, and the only instances of failure were where we didn't physically have the products."

inventory Challenge
That is a persuasive argument, certainly, but the biggest challenge facing the physical distribution side of Amazon's business appears not to come from processing shipments, which both customer surveys and third-party analysts agree works reasonably well. Instead, the bigger headache is the inventory management side of things, especially as the move into new markets gathers pace. Eager-beaver Amazon folks in the distribution centers (motivated, one hopes, by stock options that are above water, not below it) may care about their jobs a little bit more than a temporary employee of a third-party logistics provider doing the same job. Unfortunately, computer-based inventory management systems are above such things.

The ultimate model of how the system is supposed to work, as sketched out by Bezos, is a sort of one-stop shopping mall, delivered to your door in a single consignment - ideally, from a single shipping point in order to save costs. Imagine, goes the theory, being able to buy not just a video about barbecuing, and some recipe books, but the barbecue itself - and all in a single transaction.

Which is fine as theories go, but the problem that it immediately poses is one familiar to any supply-chain executive. Mathematically, the greater the number of items in an order, the greater the probability that out-of-stock items will prevent the order being shipped in one hit. Sure, there are partial fixes: higher safety-stock levels, responsive suppliers (some of whom are contracted to dispatch to Amazon within hours of an EDI order being received), extended delivery promises, faster turnaround times - the list goes on. But although all of these help ameliorate the problem, the cold hard logic remains.

And, as last year's financial results showed, Amazon's track record in inventory management can be described, at least for now, as spotty at best: markedly lower inventory turns on CDs, for example, and a surprisingly high level of inventory write-offs. Forecasting errors and seasonal factors explain a chunk of this, but some was just plain self-inflicted. "Some inventory wasn't optimally placed," Wilke concedes.

"Our vision is that if we have 20 million customers then we should have 20 million stores." - Jeff Bezos of Amazon.com

This is another area where the Manugistics software will bring improvements. Amazon will use NetWORKS to set optimal inventory levels and decide appropriate product mix and storage capacity across the network, according to a Manugistics statement. The solution will even suggest season pre-builds and optimal lane volumes to maximize customer service during the holiday season and other peak periods. By providing greater inventory visibility, NetWORKS additionally will enable Amazon to consolidate individual shipments across all shipping operations.

Moreover, the NetWORKS transportation module will simultaneously plan Amazon's inbound and inter-facility shipments for greater efficiency and better control, says Manugistics. Amazon will be able to automatically tender loads on the internet, track inbound shipments, pay carriers, allocate freight costs and run historical reports for improved transportation planning. Dynamic messaging capabilities will give Amazon and its trading partners the ability to identify and proactively solve transportation problems.

Another key to better operations is Amazon's secret weapon: the wealth of data that the company has acquired on what people buy online, how they buy it, and what combinations of things they like to buy together. Amazon executives are tight-lipped about what their plans are for this data, but a crack team of Amazon techies already is on the job, tasked with wringing as much information from it as possible. And make no mistake, the potential gains from the data are enormous.

One success story to date: the company's deployment of collaborative filtering technology, which allows it to make recommendations to customers about the kinds of products they might like. "Our vision," Bezos explains, "is that if we have 20 million customers, then we should have 20 million stores. If you never buy romance novels, then we shouldn't clutter your view with them. If you like literary fiction, then we should tilt your store toward literary fiction. The way to think about it is as a return to days of yore, when your small-town merchant knows you as a person, and knows your taste, and would say, look, I know you buy John Irving, and there's this new author who you're going to love."

Although the math is complicated, collaborative filtering is conceptually simple, he explains. "What you do is to look at all the things that an individual customer has purchased from us, and then look at all our other customers to identify those customers who have similar tastes and purchase histories," he says. "Then, in a statistical way, we merge those customers into something that you can think of as being an electronic soul mate. And then we look at the things that that electronic soul mate has purchased, but that this particular individual customer has not. And then we recommend those things."

Indeed, the terabytes of customer buying data that Amazon is collecting is a potentially unique gold mine for both Amazon and its supplier base. And, Wilke says with a grin, that explains why "the professors who are calling us, wanting to work on it with us, are who they are - we probably have one of the richest mathematical opportunities available anywhere."
Naturally, given last year's inventory management problems, slicker inventory handling is high on the list - in particular, optimizing decisions about where to store certain SKUs, and a better understanding of which items tend to be bought together, so as to facilitate co-location.

"One of the things that differentiates us from other retailers is the way we hold inventory as a network," says Wilke. "It's uneconomic on sheer cost grounds to hold all the items we stock in each of our distribution centers, so our objective is to get much better at predicting which items people buy together - which is one way we mine the data. At present, if we predict wrongly, we try and buy it into the applicable distribution center to meet a particular shipment, but if we can't do this in the timeframe that we promised the customer then we'll split the shipment. And we'll pay for the split."

Which again is where those terabytes of customer data come in. If the information that they contain is of great worth to Amazon, it's almost as valuable to the company's suppliers. Unlike some physical retailers, Amazon is happy to share the information it gathers - perhaps viewing it as a lever against any strong-arm "don't do business with Amazon" pressure on suppliers from competitors, pressure to which the company is especially vulnerable as it ramps up volumes in its new markets. Not that Amazon has evidence of such pressure, Wilke stresses, but the company is undeniably keen to make sure that manufacturers are aware of the benefits that a tie-up with Amazon offers.

Such as forecasts and customer affinity data coming from a single source, untouched by physical store-level variation. And supplied "on tap," as frequently as manufacturers want it. And cut and diced just how suppliers want it: by geography, by socioeconomic or demographic classification, or by linked purchase - customers who bought this Black & Decker drill also bought these tools and accessories.

Plus, of course, the substantial cost advantages that Amazon offers manufacturers. "They don't have to send people to maintain an end-cap display at the store, and they don't have to send their people to count and shuffle inventory. Nor do they have to ship us point-of-sale display advertising," says Wilke. "We don't want any of that, and don't need any of that - and all those things add up to hundreds of basis points to a cost structure."

Imagine, he asserts, the cost savings that stem from a Barbie doll in Amazon-friendly packaging: no point-of-sales frills - because the point-of-sale is the customer's computer screen - just the raw Barbie in a cheap 'n' cheerful cardboard tube, generating savings for both the manufacturer and Amazon. Not that Barbie-in-a-tube will be shipping from an Amazon distribution center anytime soon, he hastens to add. The analogy is an illustration of tomorrow's capability, not today's.

But it is this kind of idea, together with a commitment to global expansion and supply-chain excellence that is helping ensure there will be a tomorrow for Amazon. Bezos is certain of it, and he's a hard man to bet against.

The next six months or so are going to be the most critical in Amazon.com's short but spectacular history. By then, reckon such top Wall Street analysts as Merrill Lynch's "New Economy" guru Henry Blodget, it will become clear if Amazon.com's business model will generate sustainable profitability and positive cash flow.

The implications are far reaching. Amazon has become something of a bellwether of the New Economy as a whole. If Amazon makes it, expect the rest of the dotcom crowd to mirror its strategies. If it doesn't - well, the bloodbath will make NASDAQ's April and May decline of 35 percent or more seem like a picnic. Amazon's supply-chain strategies and its fledgling overseas expansion could well determine the outcome.

The facts are simply stated. In a little over five years, Amazon.com has captured 20 million customers in 150 countries - 10 million of them in 1999 alone. To sleepy physical world retailers - you know, those sticks-in-the-mud who still have only shops - "Amazon" has become a verb, and one that increasingly generates the kind of fear associated with the news that Wal-Mart is going to turn up on your doorstep.

For founder and CEO Jeff Bezos, it's sweet vindication. During the five years that Amazon has been rewriting the rules of retailing, he recalls, the company variously has been called Amazon.con, Amazon.toast "and my own personal favorite: Amazon.org, because we were clearly a not-for-profit company." Not bad for a quietly studious investment banker who'd headed to work each day at a desk in New York, until, in 1994, he'd spotted some statistics on the growth of something called the internet.

Yet that success has a darker side: Amazon's famous unprofitability. For all its success compared with other recent dotcom fiascoes, the fact remains that its spiraling revenue, which reached $1.6bn last year, has been matched by equally spiraling losses: $31m in 1997, $125m in 1998, and a massive $720m in 1999 - almost six times the previous year's figure. And second-quarter 2000 losses reached $317m, prompting yet another collapse in the stock price as investors fretted over slowing sales growth and rising distribution costs.

Bezos remains sanguine - if visibly more irritated with the doomsayers who write the company off. "E-commerce is a scale business, with costs that are largely fixed," he insists. And the classic solution to dealing with a high fixed-cost base? Volume - and lots of it.

So Amazon is expanding fast from its core books, music and videos beachhead. Consumers clicking onto Amazon.com these days find themselves facing an almost bewildering variety of offerings: electronics, software, toys and video games, lawn and patio products, health and beauty items, kitchenware, tools and hardware, pet care, home furnishings - some 18 million unique items with more being added all the time.

"Two years ago, we were U.S.-only, and books-only," Bezos observes. "We weren't selling music, or videos, or any of the other product categories we now sell. Two years later, although our U.S. books business has continued to grow rapidly, more than half of our revenue comes from these new categories and businesses." And with every additional dollar, the fixed costs of the distribution centers and headquarters systems are amortized that little bit further.

And make no mistake: The fixed cost base that Amazon has incurred with its distribution operation is a hefty one. In the last year, explains Jeffrey Wilke, vice president and general manager of operations, and a veteran of AlliedSignal and Andersen Consulting, the company has increased its worldwide distribution capacity from 300,000 square feet to more than 5 million square feet. The combination of rapid customer growth and rapid product line expansion, he explains, meant that Amazon's original two distribution centers - one in Seattle, one in Delaware - simply couldn't cope. The original 93,000-square-foot, largely manual facility now seems ludicrously tiny for a business of Amazon's size. Even the 202,000-square-foot facility in New Castle, Del., which doubtless appeared a brave gamble back in November 1997, now seems woefully undersized.

So four further U.S. distribution centers have opened in the last year: one in Fernley, Nev.; one in Coffeyville, Kan.; one in McDonough, Ga., and one in Campbellsville, Ky. (A further facility came with the acquisition of e-retailer Tool Crib last October, which at the time of writing is still being used solely to handle tool orders.)

Highway Automated DCs
Each of these distribution centers, says Wilke, has roughly 700,000 square feet of warehouse space, although that's an average that varies widely, as does each distribution center's provenance. Stanley Tools formerly owned the 320,000-square-foot Fernley facility, located in the Nevada Pacific Industrial Park some 30 miles east of Reno along Highway 80. On the other hand, the greenfield Coffeyville distribution center in Kansas, at 850,000 square feet is more than two and a half times the Fernley facility's size.

The distribution centers are highly automated, using barcodes and scanners to track and sort inventory that moves through the facilities. A typical Amazon DC has 10 miles of conveyor belts and ships most in-stock items within 24 hours of the order click.

But for there to be a tomorrow for Amazon - and remember the battering the company's stock price has taken in recent months, slumping from around $115 to a touch under $35 - the company has to suck in still more volume to amortize its high fixed costs. With growth in the core U.S. books and music market slowing down, and new product areas still at the fledgling stage, the sources of this extra volume are limited.

Which is why the company's management has started to expand the business geographically, as well as into new product areas. Amazon now has a U.K. subsidiary, headquartered in Slough, west of London, employing around 500 people - Amazon.co.uk - as well as a slightly smaller German one, Amazon.de, headquartered in Regensburg, Germany.

More are planned. Bezos is tight-lipped about where precisely these will be, but executives such as Wilke drop broad hints: Look at national levels of internet penetration, and overlay per capita disposable income figures, he suggests. Certainly, the establishment of a Europe-wide call center operation in The Hague suggests that further European ventures are close to completion.

Identifying desirable global locations for new distribution centers is one use Amazon will make of new supply-chain software from Manugistics of Rockville, Md. Amazon announced in May that it would install Manugistics' NetWORKS solutions to support its global expansion and operational improvement initiatives. It will use NetWORKS Strategy to model fixed and variable network costs, taking into consideration such factors as varying transportation and supplier lead times, and global constraints such as tariffs and taxes. The model will then be used to design an optimal global network.

Amazon.com's initial steps toward international expansion appear to be going well. Market research numbers from Media Metrix place Amazon.de and Amazon.co.uk as the most visited e-commerce site in their respective countries, and in the top 10 of sites visited overall in each country. But neither operation - either individually or together - will make or break Amazon on its own.

Instead, they serve as a stark test-bed. The dilemma is simply stated. To survive, Amazon must increase its volume by succeeding in performing a trick that most U.S. retailers find challenging: translating a highly successful U.S. format into one that works just as successfully overseas. And most don't. In recent years, for example, Toys 'R' Us, Wal-Mart, Staples and others have experienced disappointing results when transplanting their U.S. business model to foreign soil.

Is Bezos worried? Amazon's core offerings, he is convinced, will play as well in Peterborough as they do in Peoria. "When you move to a new country, you never find that people say, 'Could you make this [web site] a little harder to use?' It doesn't happen that way. Nor do they say, 'Can you offer less choice, less selection, and higher prices?' The core things really are the same," he asserts.

It is, of course, too soon to tell. But the signs so far are promising, according to Steven Frazier, Amazon.co.uk's managing director, despite the fact that losses as a percentage of sales are declining slowly. In The second quarter, the U.K. and German operations together lost $35m on sales of $73m, compared with $16m on sales of $31m for the same quarter last year.

Overall, the German and U.K. operations made $167m in sales in 1999, Frazier points out, up from $22m in 1998, and are currently achieving a run rate of $280m a year. Both businesses were acquired in 1998 from existing online book retailing companies - in the U.K.'s case, an outfit called BookPages - and Amazon has been careful to retain key staff and local market knowledge, he says. Despite the similarities between the Amazon.com and Amazon.co.uk web sites, "we're trying to be a local business serving a local market."

Nevertheless, there are differences. Forget the one-stop trip to the virtual mall, for example. Amazon.co.uk merely sells books, videos and music - the core Amazon.com formula. Still, for U.K. and German customers deprived of large U.S.-style physical bookstores (the vast majority of the population) Amazon's range is tantalizing: Amazon.co.uk sells 1.2 million British books in print, more than 250,000 U.S. titles, 220,000 CDs and 23,000 videos. Other products will come, but will take time to do so, Frazier concedes. "The limits to what we can do are huge, but on a day-to-day basis there's a brutal process of prioritization: What do you do next?"

For one thing, Amazon.co.uk began offering software and toys, driving sales by awarding existing customers free gift vouchers on top of hefty discounts.

Growth already has driven the U.K. operation from a small distribution facility close to the Slough HQ to a 220,000-square-foot distribution center in the Marston Gate logistics park in Milton Keynes, a hub north of London and on the country's main arterial freeway, the M1. Both operations were kept going to cope with the busy Christmas period, but then Milton Keynes took sole charge. And even this is a temporary option according to Stuart Armstrong, the company's U.K. head of operations, while a more permanent home, with a level of automation comparable to that found in the U.S., is being erected.

Are there any differences between the U.K. and the U.S. modes of operation? Although the basic philosophy - high levels of automation, for example - is the same, Europe's relatively compact nature, at least compared with the U.S., also has proved a boon. In contrast to the U.S., where the distribution model revolves around large distribution centers located equidistant from major population centers, Europe allows Amazon to play more sophisticated tunes.

Take its proximity to suppliers, for example. The Marston Gate distribution center is close to the warehouses of several book, music and video suppliers, most of who deliver twice a day, Armstrong points out. If an item is out of stock, when a customer orders it, it can be shuttled over on the next delivery, and still sent on time to the customer. And whereas 32 percent of Amazon's U.S. shipments are funneled to parcel carriers like FedEx and UPS, that is not necessary in the U.K., where the Royal Mail's standard service is 1-day delivery. "No country ever appreciates its local postal system, but the Royal Mail and Deutsche Post are among the best in the world," observes a happy Bezos, genuinely delighted to find the opportunity to offer customers an even better deal.

And in some cases, a deal that is not yet possible in the U.S. - same-day delivery, at least within London. "We began offering it through a private carrier, but then found that Royal Mail would offer the same thing," says Armstrong. So, provided that customers order within a given time window, they are offered the option of same day delivery as a free upgrade.

Positive Cash Flow
Will it, in the end, be enough? Even with U.S.-style 50 percent discounts and the rest of the Amazon customer experience? Certainly, with the share price as it is, customers are just about the only people who are feeling bullish at the moment. Employees, managers and investors alike are bruised and battered. Suppliers, too, will be heeding analysts' fears over running out of cash - fears that Amazon dismisses with flat denials and projections of a positive cash flow by quarter four.

One of the targets of Amazon's critics is its investment in distribution centers. Why incur the expense of establishing such operations at all, they ask. In a business struggling toward operating profitability, why add fixed leasing and employment costs to the cost structure when the use of third-party contractors would permit a much greater degree of variability - not to mention allow Amazon to focus more on its core specialty, online retailing?

Quite apart from the delicious irony that physical distribution is Amazon's only profit center on a straight cost basis (the 1999 accounts reveal that it charged its customers $239m for shipping their goods to them, but only incurred shipping costs of $227m), there is a strategic issue at work.

At its heart, Bezos explains, is the belief that "the end-to-end, 'click-to-ship' customer experience that Amazon creates is even more important in the online world than it is in the physical world." Why? "Online, the word-of-mouth impact is amplified," he explains. "Every internet customer has a big megaphone, and if we make a customer unhappy, they don't tell five friends, they tell 5,000. And the reverse is true: You create evangelists. If you do a great job for customers, then they tell 5,000 friends, in newsgroups and listservers and chat areas, etc. - so that online, your marketing dollars are best spent building great customer experiences."

And contrary to more recent arrivals' massive marketing and advertising budgets, Amazon's promotional expenditure does indeed remain relatively low for a business of its size - $141m in 1999, up from $60m the year before. "In our first year, we did not do a dollar of paid advertising, and we still grew incredibly rapidly," Bezos says, pointing out that in its first six months of trading, Amazon had shipped books to more than 80 countries. After a year, the number of countries shipped to had climbed to 150.

"Today," he adds, "we do a significant amount of paid advertising, but it still accounts for less than half of our new customers." Time after time, he says, when probing the reasons why new customers came to Amazon, company studies have shown that more than half the new business comes from word-of-mouth referrals.

So, goes the logic, every element of control that Amazon cedes over its operations is an element where that customer experience can be put at risk. Wilke puts it this way: "We think that it is because we control the entire process from click-to-ship that our customer experience is what it is." While the company has of course looked at third-party operators, he says, it has always come to the view that "what we do ourselves we'll do better than the market can offer right now."

In part, the issue is the creation of a sense of ownership, exemplified by Bezos's insistence that every employee in the transaction chain, however lowly, is compensated in part with stock options - and the greater a person's impact on the overall customer experience, the higher the proportion of stock options that constitutes their salary. "If you want people to behave like owners, you have to actually make them owners. There are no shortcuts to that process," he says.

For proof, notes Wilke, look no further than last Christmas - the first real e-Christmas, according to many pundits. It was, for many B2C e-commerce companies, an enormous public relations disaster, with a huge number of column inches devoted to how angry consumers found themselves without the products that they'd ordered. "If you look at those sites," says Wilke, "many of the ones who had trouble were using third-party logistics contractors. We decided to do it ourselves - and spent a great deal of money doing so - but by December 24th every distribution center was completely clean: we'd shipped 99 percent of our orders, and the only instances of failure were where we didn't physically have the products."

inventory Challenge
That is a persuasive argument, certainly, but the biggest challenge facing the physical distribution side of Amazon's business appears not to come from processing shipments, which both customer surveys and third-party analysts agree works reasonably well. Instead, the bigger headache is the inventory management side of things, especially as the move into new markets gathers pace. Eager-beaver Amazon folks in the distribution centers (motivated, one hopes, by stock options that are above water, not below it) may care about their jobs a little bit more than a temporary employee of a third-party logistics provider doing the same job. Unfortunately, computer-based inventory management systems are above such things.

The ultimate model of how the system is supposed to work, as sketched out by Bezos, is a sort of one-stop shopping mall, delivered to your door in a single consignment - ideally, from a single shipping point in order to save costs. Imagine, goes the theory, being able to buy not just a video about barbecuing, and some recipe books, but the barbecue itself - and all in a single transaction.

Which is fine as theories go, but the problem that it immediately poses is one familiar to any supply-chain executive. Mathematically, the greater the number of items in an order, the greater the probability that out-of-stock items will prevent the order being shipped in one hit. Sure, there are partial fixes: higher safety-stock levels, responsive suppliers (some of whom are contracted to dispatch to Amazon within hours of an EDI order being received), extended delivery promises, faster turnaround times - the list goes on. But although all of these help ameliorate the problem, the cold hard logic remains.

And, as last year's financial results showed, Amazon's track record in inventory management can be described, at least for now, as spotty at best: markedly lower inventory turns on CDs, for example, and a surprisingly high level of inventory write-offs. Forecasting errors and seasonal factors explain a chunk of this, but some was just plain self-inflicted. "Some inventory wasn't optimally placed," Wilke concedes.

"Our vision is that if we have 20 million customers then we should have 20 million stores." - Jeff Bezos of Amazon.com

This is another area where the Manugistics software will bring improvements. Amazon will use NetWORKS to set optimal inventory levels and decide appropriate product mix and storage capacity across the network, according to a Manugistics statement. The solution will even suggest season pre-builds and optimal lane volumes to maximize customer service during the holiday season and other peak periods. By providing greater inventory visibility, NetWORKS additionally will enable Amazon to consolidate individual shipments across all shipping operations.

Moreover, the NetWORKS transportation module will simultaneously plan Amazon's inbound and inter-facility shipments for greater efficiency and better control, says Manugistics. Amazon will be able to automatically tender loads on the internet, track inbound shipments, pay carriers, allocate freight costs and run historical reports for improved transportation planning. Dynamic messaging capabilities will give Amazon and its trading partners the ability to identify and proactively solve transportation problems.

Another key to better operations is Amazon's secret weapon: the wealth of data that the company has acquired on what people buy online, how they buy it, and what combinations of things they like to buy together. Amazon executives are tight-lipped about what their plans are for this data, but a crack team of Amazon techies already is on the job, tasked with wringing as much information from it as possible. And make no mistake, the potential gains from the data are enormous.

One success story to date: the company's deployment of collaborative filtering technology, which allows it to make recommendations to customers about the kinds of products they might like. "Our vision," Bezos explains, "is that if we have 20 million customers, then we should have 20 million stores. If you never buy romance novels, then we shouldn't clutter your view with them. If you like literary fiction, then we should tilt your store toward literary fiction. The way to think about it is as a return to days of yore, when your small-town merchant knows you as a person, and knows your taste, and would say, look, I know you buy John Irving, and there's this new author who you're going to love."

Although the math is complicated, collaborative filtering is conceptually simple, he explains. "What you do is to look at all the things that an individual customer has purchased from us, and then look at all our other customers to identify those customers who have similar tastes and purchase histories," he says. "Then, in a statistical way, we merge those customers into something that you can think of as being an electronic soul mate. And then we look at the things that that electronic soul mate has purchased, but that this particular individual customer has not. And then we recommend those things."

Indeed, the terabytes of customer buying data that Amazon is collecting is a potentially unique gold mine for both Amazon and its supplier base. And, Wilke says with a grin, that explains why "the professors who are calling us, wanting to work on it with us, are who they are - we probably have one of the richest mathematical opportunities available anywhere."
Naturally, given last year's inventory management problems, slicker inventory handling is high on the list - in particular, optimizing decisions about where to store certain SKUs, and a better understanding of which items tend to be bought together, so as to facilitate co-location.

"One of the things that differentiates us from other retailers is the way we hold inventory as a network," says Wilke. "It's uneconomic on sheer cost grounds to hold all the items we stock in each of our distribution centers, so our objective is to get much better at predicting which items people buy together - which is one way we mine the data. At present, if we predict wrongly, we try and buy it into the applicable distribution center to meet a particular shipment, but if we can't do this in the timeframe that we promised the customer then we'll split the shipment. And we'll pay for the split."

Which again is where those terabytes of customer data come in. If the information that they contain is of great worth to Amazon, it's almost as valuable to the company's suppliers. Unlike some physical retailers, Amazon is happy to share the information it gathers - perhaps viewing it as a lever against any strong-arm "don't do business with Amazon" pressure on suppliers from competitors, pressure to which the company is especially vulnerable as it ramps up volumes in its new markets. Not that Amazon has evidence of such pressure, Wilke stresses, but the company is undeniably keen to make sure that manufacturers are aware of the benefits that a tie-up with Amazon offers.

Such as forecasts and customer affinity data coming from a single source, untouched by physical store-level variation. And supplied "on tap," as frequently as manufacturers want it. And cut and diced just how suppliers want it: by geography, by socioeconomic or demographic classification, or by linked purchase - customers who bought this Black & Decker drill also bought these tools and accessories.

Plus, of course, the substantial cost advantages that Amazon offers manufacturers. "They don't have to send people to maintain an end-cap display at the store, and they don't have to send their people to count and shuffle inventory. Nor do they have to ship us point-of-sale display advertising," says Wilke. "We don't want any of that, and don't need any of that - and all those things add up to hundreds of basis points to a cost structure."

Imagine, he asserts, the cost savings that stem from a Barbie doll in Amazon-friendly packaging: no point-of-sales frills - because the point-of-sale is the customer's computer screen - just the raw Barbie in a cheap 'n' cheerful cardboard tube, generating savings for both the manufacturer and Amazon. Not that Barbie-in-a-tube will be shipping from an Amazon distribution center anytime soon, he hastens to add. The analogy is an illustration of tomorrow's capability, not today's.

But it is this kind of idea, together with a commitment to global expansion and supply-chain excellence that is helping ensure there will be a tomorrow for Amazon. Bezos is certain of it, and he's a hard man to bet against.