November 6, 2000

Printer ready

Virtual Enterprise Comes Of Age

B-to-B's compelling economics prompt many businesses to consider collaborative computing

By Andrew Binstock

More on collaborative commerce:

sidebar:Virtual Enterprise Boosts Collaborative E-Commerce Trading Partners Collaborate To Increase Sales (10/30/00) HARNESSING CORPORATE KNOWLEDGE (8/7/00) Tele.com: Easy Access, Courtesy of C&W (10/30/00)

Send Us Your Feedback

n the early 1990s, a model for business design known as the virtual corporation enjoyed a brief vogue. Under this model, a business served as the front end for a string of relationships between itself and various suppliers of services and goods. Such a company often performed some activities beyond coordination, creating added value that it alone could provide.

MIPS Technologies Inc., the maker of microprocessors that for years drove Silicon Graphics Inc.'s workstations, was such a company. It designed microprocessors that were fabricated by a different manufacturer and sold by yet other companies. Even though the chips were known as MIPS processors, MIPS did none of the hands-on work except the design--and, of course, the development of the MIPS brand. Contrast this with Intel, a company that prides itself on its vertical integration, designing, fabricating, selling, and supporting its own processors.

The MIPS model of the modern company found its blueprint in The Virtual Corporation: Structuring and Revitalizing the Corporation for the 21st Century, by William H. Davidow and Michael S. Malone (HarperBusiness, 1993). The authors depicted the virtual company as one that enabled companies to provide better products and services, since unique expertise was applied at every step in the process, and to do so more efficiently.

This model became particularly successful in manufacturing contexts, but it never gained widespread acceptance because the technology needed to integrate companies with their suppliers and customers didn't exist. Electronic data interchange was the closest most manufacturers could come to technology that could implement this model. With EDI, orders for parts typically could be sent directly to a provider's IT system, and related functions, such as checking inventory levels, could also be performed. Although EDI worked well, it was expensive and tended to be proprietary.

Ten years later, the technology for deep integration--even collaborative computing--between vendors and suppliers exists. In light of technological advances, companies are beginning to revisit the virtual-corporation concept.

The updated version is dubbed the virtual enterprise. Under this model, companies specialize in performing a handful of processes that are critical to their success, and then contract with other vendors to perform the rest. "This is not just outsourcing. Outsourcing in its early form referred to contracting out noncore, nonmission-critical services such as cafeteria services or payroll," says Michael Hammer, president of Hammer & Co., which assists companies in achieving this transformation. He adds that as managers became more comfortable with outsourcing, they farmed out other processes such as Y2K remediation and the running of routine IT tasks as well.

By contrast, in the virtual enterprise, key processes are outsourced to companies whose errors can have a huge impact on a company's performance. "This is outsourcing on steroids," Hammer says. An example of such a company is Cisco Systems, which handles product development and sales itself, but outsources manufacturing and distribution. In addition, product development on older product lines is consigned to overseas development shops that are hired by Cisco to continue supporting the products.

What drives this model is pure economics. Take the example of Adaptec Inc., a marquee name in storage components for PCs and servers. The company designs special application-specific integrated circuits (ASICs) for its products. Previously, these processors were fabricated in Singapore as needed. Turnaround time from order placement to delivery of the processors was 105 days. To shorten this, Adaptec virtualized this portion of its business, entering into a collaborative-computing arrangement with all of its chip suppliers and fabricators. The effort cost $1 million. But within months of getting the project off the ground, Adaptec had shrunk order-fulfillment time on ASIC delivery to 55 days, netting an immediate $2 million in savings. Adaptec also saves another $2 million annually in reduced personnel and inventory-carrying costs, thanks to better management of resources.

"The return on investment of converting to this model was compelling," says Dolores Marciel, Adaptec's worldwide VP of materials. "We took the risk, got it running, and the return turned out to be even greater than we expected. How often can you say that about an IT project?"

To move toward this model, companies must completely rethink the way they do IT--not just in terms of how they share information with suppliers, but also in determining how to share confidential information, as well as how they allow external vendors to drive internal IT processes that are critical to business success. "The first obstacles to implementing a virtual enterprise are organizational, not technological," Hammer says. "Technology is simply the enabler."

For example, if you contract out manufacturing and distribution, then you've put information about all your latest advances into the hands of companies that might also be building products for your rivals. The distribution company knows your customer list, and to do its job well, it must be able to access, or even drive, IT functions on your systems from its own sites. Before you can get comfortable with this view, you have to radically redefine the way your company views itself. This mind change must precede all other decisions.

Suppose an existing company finds the efficiencies of the virtual enterprise sufficiently compelling to explore. How should it proceed and how can it evaluate the impact of such an operational change on IT?

The first step, says David Linthicum, chief technology officer of Saga Software Inc. and a longtime industry analyst, is to distinguish the business-to-business issues from the problems of enterprise application integration. The latter, though challenging, represents an implementation aspect of the business-to-business strategy.

Business-to-business issues should address a company's business processes--how can these be outsourced and shared with a supply chain in a collaborative-computing context? The difficulty here is that "most companies don't understand their business processes well enough to define them or devise metrics for them, much less to outsource them in discrete and intelligent ways," says Terry Retter, director of strategic technology services at PricewaterhouseCoopers. Companies that can do so are much more able to step up to the challenges of collaborative computing.

Dell Computer, for example, has long been viewed as a pioneer in inventory management because of its ability to build PCs to order and deliver parts for those PCs on a just-in-time basis. To do this, it had to break down its processes into neatly defined subprocesses that could be individually automated. At a simplified level, for example, the assembly of a PC has eight basic steps: order processing, preproduction, making kits of parts, building, testing, boxing, delivery preparation, and delivery. Each step involves a series of smaller steps, and each is just one piece of an even larger chain of events.

Once you've identified a business process suitable for collaborative computing, the challenges of integrating applications with your supply chain begin.

The technological issues are twofold: establishing a common format for the interchange of documents and data, and integrating the applications at the various points along the supply chain. The proposed standard for Web data, the Extensible Markup Language, is widely viewed as the specification to which business-to-business documents will conform. Norton Greenfeld, principal at Implements Inc., an advisory firm for the management of sophisticated software-development projects, supports the adoption of XML. "Its great benefit is that XML contains metadata--data about the data--so that senders and recipients are able to describe to each other what they're sending and how it's laid out," Greenfeld says.

Beyond the Extensible Markup Language, there lies the need to define what data should be in the document to begin with. For example, what are the necessary data fields for a requisition for microprocessors?

To solve this problem, diverse consortia are collaborating on devising industry-by-industry document standards suitable for implementation in XML. Among these is RosettaNet, a well-known nonprofit consortium dedicated to establishing standards for electronic components. A similar effort is under way with Microsoft's BizTalk initiative, which seeks to codify XML tags for a wide variety of industries.

The need for XML and industrywide implementations exists only for companies that plan to deal with more than one partner in the supply chain. If, in fact, you collaborate with only one partner that masks its own supply chain, then you can simply arrive at an agreement with that vendor about data and document formats. "When we began our effort," says Adaptec's Marciel, "there was no XML nor any specifications for standard interchange of this kind of information. So we just rolled our own, and it worked out well."

Greenfeld points out that this model works as long as the roster of partners stays the same. However, if Adaptec brings other foundries into its supply chain, it will likely have to retool this portion of its collaborative-computing environment to support any existing industry standards.

The simple interchange of documents isn't enough. Managers also need to be able to drive processes of their supply-chain partners. For example, customers want to be able to generate inventory reports on parts held by a vendor. A vendor wants to be able to create invoices at the source company for the products it ships on the company's behalf. Running these processes across the Internet means unprecedented technological access to back-end IT systems.

"This is the really hard part," says Saga Software's Linthicum, because this is the point at which enterprise application integration becomes the critical technology.

Many companies specialize in EAI technology, but only some of them work in the area of supply-chain integration. Two notable companies that focus on supply-chain integration are Extricity Inc. and Netfish Technologies Inc.

In essence, these companies provide the software that permits the integration of in-house IT systems (specifically, existing back-end systems and most databases), as well as the front-end connection to the supply chain. Other technologies--from Actional, Saga Software, Vitria, and others--help integrate these functions within the greater enterprise. They might, for example, enable customer-relationship-management packages to access supply-chain information.

Dennis Byron, director of enterprise-application research for market researcher International Data Corp., foresees a time when this market will change as the ERP market leaders begin making moves. "You expect many manufacturers to have SAP running on their shop floors and in the back office. And SAP started rolling out its supply-chain software two years ago," Byron says. "Soon SAP will need to provide its own EAI technology to support the virtual enterprise. We expect that SAP will do this through acquisitions."

Byron says that as with other ERP vendors, such as J.D. Edwards, Lawson, and PeopleSoft, SAP will enhance its offerings by acquiring or bundling integration technologies from other companies.

For IT shops running their own software rather than packaged back-end applications, significant customization will be necessary. Programming interfaces to the applications will have to be developed, published, and tested, and data needs to be accessible in formats that partners will recognize. "Great care should be extended in designing the APIs, because once they're in use, they can't be changed without having to change software at every vendor in the chain," Implements' Greenfeld says.

In addition, he notes that data formats for interchange should conform to the XML frameworks for each industry. If there's no framework available, companies must make sure to export at least some sort of XML, so partners can quickly tailor their applications to meet an individual company's needs.

Byron predicts that most ERP packages will move to less-monolithic models, where applications work as a series of objectlike modules with published APIs and document standards. By moving in this direction, ERP vendors will make it easier for customers to build their own external interfaces to the systems.

While the customization of home systems requires careful design and implementation, this is by no means a unique predicament. Forward-thinking companies that were optimizing their supply chains before the advent of many of the technologies discussed here were forced to build the entire infrastructure by hand. Wal-Mart Stores Inc. and Cisco Systems are two companies that did so successfully (for more on Wal-Mart's most recent development plans, see "Retail's Super Supply Chains," Oct. 16, p. 22; informationweek.com/808/ walmart.htm).

Another company, Arrow Electronics in New York, the largest distributor of electronic components in the United States, sees supply-chain optimization as a critical competitive advantage. Because of this point of view, the company has been at the forefront of many technological developments in this area.

"We built an EDI infrastructure with 600 of our partners," says Lauren Holmes, Arrow's VP of global systems development. "With this EDI system, we were able to handle orders very efficiently; this was an advantage when wooing customers."

Now, with supply-chain optimization, Arrow can let those same customers check the status of their orders by themselves and even check directly with Arrow's suppliers as to levels of inventory for large jobs they may have. "This ability distinguishes us from many of our competitors--although, with today's fast-moving technology, we'll have to keep investing to maintain this lead over our competitors," Holmes says.

The supply chain as described in this model is still comparatively static. While the various participants are heavily interlinked at an application level, there's no standing provision for swapping vendors in and out of the supply chain in response to market conditions. As IDC's Byron points out, once an industry has a standard framework for document exchange and standard application-interface definitions, there's no reason why one vendor in the supply chain (a contract manufacturer, for example) can't easily be exchanged with another. Once that happens, the supply chain takes on a new kind of flexibility.

"This is already happening in the electronic marketplaces and in E-procurement, where materials are being bought in a competitive environment," Byron says. "Manufacturing services such as materials requirement planning, product data management, and work-order generation are likely to follow." (For more on expanding the capabilities of electronic marketplaces, see "A New Spin on EAI").

Indeed, beyond simple cost issues, additional factors, such as the location of the assembly plant, might make a given vendor desirable for one order yet unsuitable for another. By creating dynamic supply chains, every bit of pricing can be wrung out of the manufacturing process.

To PricewaterhouseCoopers' Retter, however, pricing pressure is the very reason this model won't work. While contract manufacturers and assemblers will indeed compete for business, they will quickly come to compete on things other than price, Retter says.

Value-added services will be the differentiator, much as discount stockbrokers no longer compete on price in the business-to-consumer market. As a result, companies will gravitate toward supply-chain vendors that reflect a company's core value proposition. For example, if a company's differentiator is quality of implementation, it will tend to prefer high-end vendors, at the cost of less favorable pricing.

Likewise, companies that compete on price will veer toward vendors offering the least-expensive services. Once the pool of compatible vendors is identified, companies are likely to do business with the one or two with which they are most comfortable. In essence, the supply chain will remain flexible, but not aggressively dynamic.

Not many companies today have moved to the virtual-enterprise model; the enabling technology is still very new. However, forward-looking companies that view IT as a competitive advantage are actively exploring this model. They have begun to pilot collaborative-computing projects, and they hope to apply the lessons learned from these pilots to the optimization of their supply chains.

As already noted, the return on investment can be compelling. Gary Lapidus, an analyst at Goldman Sachs, estimates that if U.S. carmakers optimized their supply chains with the technologies discussed here, they would save an average of $3,650 per automobile. Numbers like these have a lot of industries scrambling.

Andrew Binstock is the principal analyst at Pacific Data Works LLC, a firm specializing in market analysis and the composition of technical white papers. He can be reached at abinstock@pacificdataworks.com

Photos of Marciel and Retter by Brian Long
Illustration by Martin O'Neill

Back to This Week's Issue
Send Us Your Feedback
Top of the Page