29 Feb 2000  Research & Ideas

Whence IT Value?

Spending on information technology on the part of U.S. manufacturers is finally starting to pay off in increased productivity. Why now? Have IT investments, growing steadily since the 1970s, suddenly crossed a magic threshhold? HBS Professor Andrew McAfee believes the answer lies neither in magic nor in the growing power of computers themselves. Productivity gains, he writes in this article from the online journal Exec, may have more to do with the evolution of computing from PC "islands" to integrated networks that bridge distances and bring people together.


During the past few years inventory turns among U.S. manufacturers have climbed steadily, and it appears as if productivity has improved nicely. One explanation for these happy trends is that the massive investments we've been making in information technology are finally starting to pay off in visible, important ways. This hypothesis is reinforced by a growing number of academic studies indicating that a dollar spent on IT returns at least as much as a dollar spent on other forms of capital in many industries.

These findings raise an interesting question: Why is IT apparently making such a big difference now? After all, IT has been inside most organizations, especially big ones, for well over a generation. And while it is true that IT investments have been skyrocketing recently, they've been growing steadily since the mid-1970s. Did they suddenly cross some magic threshold, before which they were unproductive and after which they started carrying their weight? That seems unlikely.

Observers of this phenomenon have advanced three more-plausible explanations about the recent confluence of IT and productivity. The first is that the incredible recent advances in processing power, storage capacity, bandwidth and memory — especially on a per-dollar basis — are driving the observed business benefits. However, all of these have been getting better, faster and cheaper at about the same rate throughout the history of computing. This brings us again to a magic-threshold hypothesis.

Another explanation is that modern IT allows companies and coalitions to explore radically different organizational forms — such as so-called virtual corporations — and build whichever forms are most productive. Still, what is it about IT that is allowing this radical reconfiguration to take place now?

A final explanation is intriguing. It states that new inventions, even manifestly useful ones such as computers, require some "percolation" time before they are productively employed. Studies of the history of steam and electric power suggest that this percolation time is about 25 years. This fits pretty well with what we've observed about IT.

While the percolation argument makes sense and is appealing, I believe there's something else going on. The jumps in inventory turns and output line up very nicely with the era of networked computing. Remember, PCs were initially islands, and we started to get good at interconnecting them (outside universities and research labs) only about 10 to 12 years ago. Of course, we've become a lot better at it over the past five years, with the explosion of the Internet, intranets, extranets and other offspring of the Internet Protocol (IP).

Networks connect people to each other and to potentially vast pools of information. They are thus extraordinarily valuable tools for businesses. It may well be that businesses benefit primarily not from the computer's ability to execute algorithms but from its ability to interconnect information and the people who use it.

My own research on enterprise IT systems reinforces this conviction. I've found that the people who use networked systems in their jobs like them, even though they usually are not involved in designing, buying or implementing them. Normally, imposing any kind of technology on workers is a recipe for disaster. Why does this not seem to be true with enterprise systems? Perhaps it is because their massive, centralized databases and embedded business processes provide another means to interconnect people within the company who were previously isolated from each other and, especially, from each other's information. This is a potentially huge benefit and can overcome many kinds of cost.

Enterprise systems are not always implemented or used wisely, and they are not what every organization needs. But most organizations do need to ask themselves, "How can I put my employees in touch with the people and information they need?" When the organization identifies and implements the appropriate answer, it joins the network era and may get to join in its benefits.

Unfortunately, this is not the question a lot of organizations ask when they buy IT, nor is it the question around which many IT vendors design and sell their products. Instead, functionality continues to be king. New releases of everything from ERP to office automation software tout the new functionality they provide — in other words, the new data-processing algorithms that are contained in the software. And many hardware vendors trumpet their ability to execute these algorithms or store their output quicker, faster and cheaper.

New algorithms are seductive in exactly the same way features on a new car are seductive. But let's not forget that a car's real function is to cross distances and bring people to each other. And let's start to consider that the real function of IT might be strikingly similar.

Republished from the October 1999 issue of Exec: The Executive's Guide to Electronic Business Solutions. Copyright 1999. Used by permission.

Andrew McAfee is an Assistant Professor at Harvard Business School.