From the beginning, the Internet has sparked the imagination of dreamers and conjured up visions of a technology that would transform society. To the army of geeks and programmers who were its early proponents, the new open technology seemed to offer limitless possibilities - borderless free markets, a close-to-costless way of making transactions, and a fluid and effortless medium for collaboration.

But for this new technology to actually transform the economy, the biggest impact would have to occur in the relatively mundane arena of business-to-business transactions. Business purchases account for more than 70 percent of total sales in the economy, according to Princeton economist Alan Blinder. And more so than individual buyers, company purchasing departments are likely to have access to computers and other Web technology, as well as the incentive and mandate to cut costs.

In 1999, corporate America was finally won over. Jack Welch, the hardheaded CEO of General Electric, announced to the world and his 340,000 employees that henceforth the Internet was priority "number one, two, three, and four.” As the largest and most profitable "old economy” company in America, GE sells products ranging from aircraft engines to light bulbs to financial services to sitcoms (it owns NBC), and has revenues of $112 billion. With its enormous network of suppliers and partners, any effort by GE could influence how business is done in a wide range of other companies. And Welch, who made his reputation as a tough restructuring artist during the 1980s, is practically an icon of modern management, one of the most written about and imitated corporate chiefs of the post-World War II era.

A blizzard of B2B announcements followed Ford, General Motors, and DaimlerChrysler announced they had combined forces to develop a giant on-line exchange that would centralize suppliers to the auto industry. Metals, aircraft, medical supplies, printing companies, food and beverage supplies, construction materials, chemicals, commercial real estate, and even lawyers have all mounted sites or announced their intentions. Some are old-line companies wanting to move their own buying and selling onto the Internet; others are newer ventures offering software and services for creating and hosting new on-line exchanges. In many instances, details about the products they would offer and who would operate the sites were left vague; indeed, some cynics speculated that high stock market valuations for Internet companies, not fundamental business reasons, were the driving force behind at least some of the announcements.

Much of the subsequent press attention focused on handicapping the horse race. Whose "business model” would be the most profitable? Sites where companies post requests and exchange information but negotiate deals off-line? Or those featuring open competition and on-line auctions? Would "new economy” providers of electronic catalogs and exchange services such as Ariba, Commerce One, and VerticalNet remain in front, or would the old industrial stalwarts, such as GE and Ford, overtake them?

But trying to pick individual winners or losers so early in the race is a fool's game. There are no comprehensive and official numbers produced (the federal government won't release its first official B2B estimates until early 2001), so right now it is difficult even to gauge the extent of total B2B activity. Estimates by consulting firms put last year's B2B sales in the United States at somewhere between $50 billion and $150 billion (about three-quarters of the world's B2B sales), depending on the methodology used and what is included - transactions completed on the Internet or those merely initiated. By any account, this number is relatively small when compared, for example, to Ford's 1999 purchasing budget of $83 billion. Much of the current activity involves used equipment, excess inventory, and spot commodity products, and sales remain concentrated in computer hardware, software and data networking, and electronics, where comfort with technology is widespread. One of the top non-tech sites, W.W. Grainger, which sells industrial equipment such as motors, cleaners, and mops, had 1999 Web sales of only about $102 million out of total firm sales of $4.5 billion, and in an industry with sales estimated at more than $300 billion.

Nevertheless, the long-run potential is there. Apart from the individual winners and losers, the Internet may create real efficiencies for the economy as a whole. For some guidance as to where these efficiencies might come, consider the driving forces behind the last twenty years of financial innovation. Nobel prize-winning economist Robert Merton identifies the source of the efficiencies in three places. The first is in the creation of new and bigger markets, which can lead to lower prices and higher quality because firms can now complete transactions with firms that were either too far away or in markets that were too small to be profitable. The second is reducing the costs of making those transactions. The third is a bit more subtle, but arises because each party to any business deal usually has inside information about his or her own company. This asymmetrical knowledge tends to reduce trust between the parties and increase the costs of monitoring subsequent performance. If the Web really were to promote the sharing of information and to make the boundaries between firms less distinct, it would shrink the costs that are driven by distrust.

If only getting these efficiencies were as easy and effortless as in our imagination.

CREATING NEW MARKETS

In 1994, when GE hired Glen Meakem straight out of Harvard Business School, he decided to enter the unglamorous field of purchasing, he told Fortune magazine. The second day on the job, he took part in a conference call in which a member of GE's transportation division described a live auction that GE had arranged in a local hotel. Suppliers were able to look at samples and drawings for components that GE wanted to order, and GE managers used flipcharts where suppliers could write down bids. The scene was chaotic but prices fell to levels that amazed even GE. Inspired by this, Meakem suggested that GE create the world's first commodity market for industrial supplies. Although the first browser was still a year away, GE already had electronic data interchange (see sidebar), a sophisticated private network that linked it with its suppliers, and Meakem proposed using this for weekly auctions. His bosses at GE were interested but cautious, so they proceeded slowly. Frustrated, Meakem eventually quit GE and started FreeMarkets, an Internet auction company in the heart of the old economy city of Pittsburgh, Pennsylvania.

With its relatively low setup cost and open technology, the Internet has the power to bring many more buyers and sellers into a market, even pulling firms in from across national borders. Firms gain access to more and better suppliers, not just local companies or firms that they already know. This bigger market is also likely to intensify competition among suppliers and reduce the power of local monopoly, giving buyers access to goods at lower prices or higher quality. Bigger, thicker markets also increase the rewards and incentives for introducing new products; an idea that might not be profitable in a small market with few buyers might be very profitable if the potential market were larger.

So fragmented industries such as fruits and vegetables (ProduceOn-line.com and BuyProduce.com) and fish (gofish.com) are likely candidates for the efficiencies created by new larger on-line markets. So are standardized off-the-shelf products, like carbon steel and plastic resins, and basic commodities. (Auctions for commodities, such as oil and wheat, have already existed for some time, as futures exchanges.) For example, MetalSite and its competitor e-Steel allow buyers to easily check availability, price, and delivery date for a half-dozen or more steel producers. Since producers often make more steel than is needed to fill an order, much of what is bought on-line is excess and secondary steel, which accounts for 15 percent of the steel sold in the United States and would previously have sat in warehouses for much longer.

But creating a new market for specialized and complicated manufactured products, such as plastic injection moldings and metal castings, is more difficult. Yet, that is what FreeMarkets aimed to do. Meakem realized that the key was not simply in electronic auction technology, but also in making the effort to round up the suppliers and create product standards so that all bidders would be bidding on exactly the same job. So before any auction takes place, FreeMarkets works with clients to create elaborate specifications that completely spell out the product. FreeMarkets also helps locate, research, and screen a pool of competitive suppliers, checking on each one's finances, quality ratings, and even equipment condition, before choosing who will be allowed to bid. Only then, does it hold an elaborate on-line auction that can last for up to three hours. FreeMarkets' biggest client, United Technologies, headquartered in Hartford, Connecticut, says that in 1998 it spent 43 percent less than expected for printed circuit boards used in elevators and air conditioners because of the auctions. Other customers claimed savings of between 7 and 14 percent. The firm's second biggest client, GM, reduced costs by $120 million on the purchases it made this way, according to FreeMarkets, before it left to join forces on-line with Ford and DaimlerChrysler.

FreeMarkets is not alone in trying to create new on-line markets for unorthodox, nonstandard products and services. A lawyer recently announced the launch of ELawForum, where prospective clients will package legal work and solicit bids from law firms in an on-line auction. Like FreeMarkets, law firms will be prescreened either by the client or by ELawForum prior to the bidding. During this process, the law firms can clear up questions about conflicts of interest and must specify everything, from which attorneys will be assigned to how much photocopying will cost, to whether associates will stay in first-class hotels.

But some products will be too complicated or depend too much on personal relationships, not rock-bottom prices, for the full on-line treatment. ELawForum does not expect clients to put such high-stakes matters as hostile takeovers or defending against criminal indictments up for on-line bid, at least not early on. And Goldman Sachs, in announcing it is financing the first major Web-based commercial real estate leasing and sales site, agrees that brokers may still be needed to help negotiate terms in the more complicated deals. Industry observers have also noted that on-line auctions can create a false sense of urgency that is not helpful when buying certain goods, such as used equipment; in those instances, they expect sites to facilitate initial contact but that negotiations will take place off-line. In other businesses, such as construction and manufacturing, some firms will continue to resist buying from a vendor they don't know and trust because the cost of shutting down operations in the event the vendor doesn't deliver is just too high.

There are also legal and other limits - trade laws, tariffs, quotas, and government regulations - that may make it difficult to create a single national or global market. For example, U.S. steelmakers have been successful in staving off foreign competitors with trade regulations and dumping charges; for now, there are tariffs on the sale of steel between industrialized countries (they are expected to be dropped in 2004). Thus, a buyer and seller from two different countries have to figure out such questions as whether the type of steel product is under quota; the amount, if any, of tax or tariff; and whether there are any other applicable regulations from either country. So while buyers and sellers can meet on MetalSite and e-Steel, they have to conduct negotiations off-line, by fax, phone, or in person. And domestically, the Federal Trade Commission is reportedly looking into the proposed joint Ford, GM, and DaimlerChrysler exchange to determine if its structure is likely to lead to unlawful price signaling or coordination among buyers or sellers.

CUTTING THE PAPER TRAIL

For some firms, the primary appeal of the Web is not so much in creating new markets, but in reducing the costs of exchanging information and making transactions in the markets that already exist. These costs can be considerable. Firms need to forecast demand, plan production, and alert suppliers. Later, they need to monitor sales and warn suppliers as to any changes in future orders.

Meanwhile, suppliers have to plan and schedule production and arrange for delivery, and their own chains of suppliers have to do the same. In many industries, whether a firm is a buyer or supplier depends on the context; two firms may supply each other with certain products and compete against one another in selling others, forming a dense network of buying and selling relationships among a large number of companies.

At every stage along the way and for every transaction, a tremendous amount of information needs to be coordinated, updated, and communicated. The process can entail boxes and boxes of documents catalogs, specifications, approval forms, order forms, shipping notifications, invoices and billing, remittances and payment information, and inventory and management records. Traditional paper-based systems are slow to search, costly to update, and prone to error. Electronic data interchange (EDI) networks, which connect buyers and suppliers, were first developed in the late 1960s and reduced many of these problems. Once entered on a computer, information can be moved quickly, checked for errors, updated, and shared by everyone on the network. But EDI is costly to implement and, as a private proprietary system, it makes it difficult for firms that are not a part of the same network to connect to one another.

The Internet offers firms a cheaper and easier way to connect, making EDI's efficiencies available to smaller, less technically-sophisticated firms. It also allows information to be sent at any moment, offering the opportunity for real-time collaboration. Browser-based and open in nature, Web-based networks let firms work together, even if they don't have long-term commitments to each other or a common set of computer services. This creates potential efficiencies in undertakings that involve infrequent collaborators, such as large-scale, commercial construction projects.

Blueline, of Palo Alto, California, for example, will create a Web site that connects the dozens of architects, engineers, contractors, subcontractors, project managers - all the way down to cement pourers - working together on a particular construction project to a common set of schedules, drawings, and other documents. There is no need to fax or send any pieces of paper (which in the past could take up to five days if the project was in a faraway location such as Thailand). Approvals and revisions can be made electronically and tracked easily. Time is saved and errors are reduced.

In a recent Wall Street Journal article, a project supervisor for the new plaza being built along San Francisco's waterfront opens his e-mail and finds a contractor's query about a manhole he encountered that isn't on the drawings. Should it be pulled out or incorporated into the new design? In this case, the supervisor uses the Web to figure out who should answer the question and passes it along with the necessary drawings. There is little time lost, and no problem of blurry or lost faxes. When it is decided that the manhole cover would be too difficult to move, drawings can be revised and made available to everyone involved; and a problem that could have taken days to resolve is ended in a much shorter time and without any confusion.

The Internet also offers potential efficiencies in industries where inventory costs are a major share of total costs, such as the auto industry and other retailers. If more firms can use the Web to reduce the time taken to transfer sales information and get purchase approvals, they can tie up fewer resources in inventory, reduce interest paid, and save on storage costs - all without increasing the risk of running out of stock or interrupting production because of parts shortages. To this end, some firms have even given suppliers direct access to sales data, forecasting software, and product specifications, and have made the supplier responsible for monitoring inventory and reordering when stocks fall below a certain level. Although EDI and other just-in-time inventory management techniques have already achieved significant efficiencies of this sort, the Internet will expand the possibilities.

But savings attributable specifically to the Internet are hard to calculate. One large firm recently said that it expected the costs of financing inventory to fall by half, but such costs represented less than 2 percent of the company's total costs. In the U.S. economy as a whole, business inventories, at $1.1 trillion, represent at most 20 percent of total costs (given wage and salary disbursements at $4.5 trillion), so the potential savings from reducing inventories while real are not staggering. 

And, there are various other hurdles before Web-based supply networks become widespread. For one, old habits die hard. Some project managers in big construction projects remain wary of the lack of a paper trail and insist on printing out documents anyway. Even today not all subcontractors have computers with modems, and a few still resist using computers at all. Another difficulty standards for Extensible Markup Language, or XML, the language companies use to catalog and tag data to be exchanged over the Web, are still being developed. Currently a number of different industries are working on definitions of documents and other data elements, with some observers saying that convergence to a single, well-defined standard across all industries is five to ten years away.

Security on the Internet is also an issue, though firms are perhaps less nervous than several years ago. The open nature of the Internet makes it vulnerable to unauthorized access, security breaches, and data corruption; the distributed nature of the Internet means no one is responsible for messages arriving intact and on time. Several companies offer a variety of Internet security tools, generally consisting of concentric firewalls with layers of encryption along with digital certificate systems, which allow authorized employees of suppliers to access only those parts of a site deemed necessary to their job. Yet, even off-the-shelf firewall products generally need to be customized; configuring and maintaining firewalls and other access restrictions takes both money and human resources. And a firm's security is only as strong as its weakest partner, so some firms have developed protocols and monitoring systems for the firms with which they do business, but at a cost.

And while the Internet may be considerably less expensive than EDI, that doesn't mean it's always cheap. Although a simple Web-based system is relatively inexpensive, the cost of a more elaborate system for a large firm can reach tens of millions of dollars. Splicing the Web to firm's legacy systems is also not without expense or pain - although this is easier than it used to be - with many companies continuing to use both the Web and EDI.

All these factors make it difficult to generalize about how much the Web can reduce the costs of moving information and making transactions. One measure is the reduction in purchasing agents and middle managers (for the same volume of output), but determining this is not always easy. For example, Cisco Systems, maker of routers and switches, is certain that it has achieved efficiencies by this standard. But, it also admits that its employment numbers do not offer proof because of added employees in research and development and marketing over the period. Estimates by other companies and studies by consulting firms find savings ranging from 25 percent to 95 percent compared with manual purchase orders.

THE GAINS FROM TRUST

As the Internet makes it easier and cheaper for firms to exchange information, some firms will develop more porous boundaries with their suppliers and their business customers, sharing more information and even giving over responsibility for reorders. With this reduction in private, "inside” information may come another source of savings. Fewer resources may have to go into policing and monitoring the behavior of suppliers and partners, leaving more resources available to fix problems.

Although such savings may be hard to quantify, arrangements that favor, as one company put it, "opening the kimono,” can ultimately reduce costs. For example, Dell Computer has created a customized Web page where employees of its largest suppliers can log on to a secure site. There, they can view Dell's demand forecasts and other sensitive information, such as the identity of its customers and how much equipment each is ordering. Dell also passes on data about its defect rates and requires suppliers to share information about their own quality problems. Not only can errors get fixed more quickly, but the extra time and expense that arise from distrust can also be reduced.

Yet, some firms will choose not to engage in the cooperation and sharing of information that is necessary for these efficiencies to be achieved. Many firms fear - with some justification - that giving others access to such information can turn into a competitive disadvantage. Suppliers (who are often also potential competitors) may be able to glean information that will allow them to assess a firm's strengths and weaknesses, and even steal their customers. And larger firms that already enjoy the advantages of size and muscle may oppose transferring that strength to others. Retail giant Walmart has so far resisted joining an exchange, reportedly because it doesn't want to link up with and help competitors.

It is also unlikely that firms will grant such access to firms they don't already know. Research shows that humans tend to develop trust in face-to-face situations, from vocal intonation and body language. In that sense, the Internet may prove to be a relationship enhancer - a way to connect to suppliers and customers one already knows - functioning in addition to, not in place of, face-to-face meetings.

ARE WE TRANSFORMED YET?

In many ways, the early Internet visionaries were right. Even now, the Internet can sometimes seem like magic, putting information at your fingertips, allowing easy collaboration, and reducing geographic distance to a click of a mouse. Giga Information Group Inc. estimates that the global cost savings from B2B will rise from $17.6 billion in 1998 to $1.25 trillion in 2002, with about half the projected savings going to U.S. companies. Investment bank Dresdner Kleinwort Benson figures that companies that use big on-line exchanges could slash procurement costs by as much as 10 percent.

But if the revolution is coming, it isn't here yet. A recent study by Forrester Research, of Cambridge, Massachusetts, found that only about one-quarter of the firms they surveyed will actually use the Internet for any B2B transactions this year. And based on an informal survey of 30 of the largest U.S. companies, Barron"'calculates that profits of Dow companies would increase by at most 1 to 2 percent over the next few years, with many firms still moving relatively slowly to embrace the Web. Moving around on the Internet may be effortless, but getting there requires work.

Nonetheless, we are only at the beginning of the changes ahead. Goldman Sachs estimates that 10 percent of all U.S. B2B sales will be Internet based by 2004, and other consulting firms put the number even higher. Robert Litan, director of economic studies at the Brookings Institution, told The New York Times, "My instinct is whatever we find the impact to be, it will be a lot greater five years from now.” Perhaps in the end, the Internet will prove not to have been a revolution, but a transformation that happened bit by bit.

Before B2B

The Internet is only the latest in the long line of innovations in information technology that began well before the typewriter and includes the telephone, photocopy machine, and fax, among others. Almost everyone who has written on the subject seems to have a personal favorite. Princeton economist Alan Blinder likes the laying of the transatlantic cable, completed in 1866 after several unsuccessful attempts, because it reduced the time it took to send a message from New York to London from about a week to a few minutes. Malcolm Gladwell, writing in The New Yorker, favors the bar code, because it facilitated the development of modern warehouses where products could be efficiently stored and retrieved, and it greatly reduced the necessary time and cost for products to physically make the journey from the supplier"s warehouse to the manufacturer to the retailer. It was also key to the Internet's immediate predecessor, electronic data interchange (EDI) networks.

First introduced in the late 1960s, EDI networks are private proprietary networks that allow buyers and sellers to transfer information generally using centralized electronic mailboxes and standard message format. (Point-to-point networks are less common because they require the receiver's computer to be in receiving mode whenever the sender releases information.) The network operator establishes a common language and is responsible for maintaining security tools that can scramble data, detect tampering in transit, and allow verification of the information sender's identity. Firms in the network pay a subscription fee and maintain software that translates information from their home language into the common language.

EDI was a huge improvement over paper-based purchasing systems. It greatly increases the speed and accuracy of communications between firms. It reduces the high cost of transferring information from paper to computers, and it improves firms' ability to manage inventories. Sharing information at a separate site also reduces the risk of security problems and data corruption. But EDI networks also have limitations. They tend to be batch oriented and thus not useful for real-time production, procurement, and pricing applications (such as auctions); they impose relatively hefty costs and capital investment; they also require that firms license and maintain complex software, such as the translator used to convert information back and forth to the standard format.

So, despite its advantages, EDI was never cheap enough or open enough to become ubiquitous. It was more common among large and technically savvy firms, and most common in industries where a powerful buyer pushed its suppliers to sign on. Firms that joined a network often found that the benefits were limited because other firms with which they did business were on a different one; some network operators attempted to link up, but it was not always clear who was responsible for the fees. So many firms continued to use paper, which only further constrained the potential benefits to those considering an investment in EDI. Only when it becomes easy and cheap enough for a majority of firms to be electronically linked can firms completely do away with their paper-based systems.