I have just written a book - Realising the Promise of Enterprise Systems (Harvard Business School Press, Feb 2000) - on how organisations can use and get value from enterprise resource planning, or ERP, systems. In the course of publicising this tome, I've come across what seems to me a bizarre, if somewhat implicitly articulated, idea: we don't need big enterprise systems any more, because we've got the Internet now.

Of course, I could chalk it all up to the usual retro-amnesia we encounter when any exciting new technology comes along. When PCs and client/server systems emerged, everyone predicted the death of the mainframe. Yet big-iron systems still beat away at the heart of the vast majority of large organisations, and IBM maintains a tidy business in selling and supporting these supposed dinosaurs. Similarly, fibre optics hasn't made copper wire extinct, and object-oriented programming has hardly done away with conventional programming approaches. The new technologies get all the press and are usually accompanied by revolutionary rhetoric: "Free your organisation from the tyranny of the (fill in the blank)." Somehow, however, the old stuff lumbers along and is sometimes even rehabilitated. (Mainframes don't seem nearly so oppressive any more, do they?) So before dismissing chants of "ERP is dead, long live the Internet", I engaged in some serious, if not totally objective, thinking about whether they contained any truth. Is ERP still the business necessity it became in the 1990s? Is there something about the Internet that invalidates the premise of integrated applications? In short, no. However, since I have a vested interest (book royalties) in arriving at that particular answer, you might want to read on to check my logic.

My favourite analogy to the world of IT has always been plumbing. Information is water, telecom is pipes, computers and software are valves and fixtures. If you ask me, the Internet is more pipes than anything else - fundamentally a distribution medium for information. Of course, the information that flows through the Internet is getting more amenable to processing all the time, with advances like Java and XML. But it's still processed on computers by applications, and I would argue that the best business applications around are still the enterprise packages offered by major vendors.

Of course, there are exceptions for point-specific applications. In procurement, for example, you could plausibly argue that the e-commerce-driven applications from companies like Ariba and Commerce One are as good as, if not better than, anything you can get from a mainstream enterprise system vendor. This is particularly true if your culture values autonomy and decentralised purchasing over centralised control. But there are two problems with having the argument in the first place. One is that these e-procurement vendors have alliances with enterprise vendors and are reasonably happy to allow their customers to interface their systems with the broader applications suites that pull all the different kinds of information together. The deeper, underlying problem with point-specific e-commerce applications is that they will never - at least not in the next decade or so - be able to achieve information integration within companies.

Forced Integration

The key to the value of enterprise applications is the integration they force on companies' information and processes. Sure, it seems tyrannical while you are going through implementation, but it is worth the trouble to have key information mean the same thing throughout the business, and to have key processes performed in a similar fashion. The integration allows companies to offer services like available-to-promise inventory, which means that the customer doesn't order something that can't be delivered on time. If your procurement systems aren't integrated with your sales, manufacturing and logistical systems, available-to-promise is just a tantalising mirage. A lot of e-commerce companies, despite their highly Web-enabled applications, just can't do it.

I don't see how an organisation that relies on Internet-enabled applications can achieve integration of this type. If you've got an e-procurement system from one vendor, an e-HR system from another and an e-finance application from a third supplier, how are you ever going to determine, for example, the current value of your assets? Perhaps you could get all of these different applications to talk to each other, but doing so is about as likely as integrating the "best of breed" applications was a few years ago.

If you're heavily into Internet ideology, you might retort that in the e-economy it's more important to integrate with external e-business communities than with your stodgy colleagues in finance. There's no evidence for this, but you'll make a much better partner with other companies if you can back up your commitments to them by delivering on time, shipping all components at once and selling only what's profitable.

Portal Versus Portal

Enterprise software vendors, of course, have realised that they need to get with the program on interfacing their offerings to the Internet. All have now achieved full Internet compatibility from a marketing perspective, and some have even begun to deliver software functionality. If they can move quickly enough to placate restless customers, they've got a built-in advantage: it's easier to build networks of companies that have the same enterprise software. If I've already installed the same system as most other companies in my industry, it means I already have highly similar information and processes to those other companies. Agreeing with these companies on standards and connecting with them to do e-commerce will be a lot easier than with companies that haven't gone through an enterprise systems implementation.

SAP AG recently announced, for example, that five German chemical companies had agreed to use the vendor's Internet capabilities (broadly known as mySAP.com) to form an e-commerce network with five of their suppliers. All of the chemical companies have already installed R/3, SAP's enterprise applications package. Look for more of this sort of thing in the future from industries in which several companies have adopted the same enterprise package.

I think we can also expect to see many e-commerce companies beginning to install enterprise application packages. The first priority of e-commerce companies may be to advertise on television to get some customers. Their second priority will be to treat their customers right. In order to do that, they're going to need the same information and process integration that their brick-and-mortar forebears have been struggling with for decades. Some things change, some things don't.

Thomas H Davenport is a professor of management information systems at Boston University School of Management, director of the Andersen Consulting Institute for Strategic Change and a distinguished scholar in residence at Babson College. He welcomes reader comments at davenport@cio.com