Cultural differences between large and small companies may also account for the relative difficulty large companies have in using computers to their advantage. New technology pays off especially well when people are prepared to change their work habits around it. But behavior at larger companies is often so hierarchical and ingrained that it prevents that sort of worker flexibility. There's a story about GE chairman and CEO James Welsh, who once chewed out managers for spending too much time polishing their E-mail notes, thus wasting the speed and interactivity computer networks can promote. How did the managers respond? Well, they continued to polish their notes, but then they took the time to go back and insert typos and other mistakes to give their E-mail that dashed-off quality Welsh was after.
But where large companies are most at a disadvantage is in the nature of the technology itself. Big companies' biggest problem? They're stuck with mainframes.
It's not a question of hardware cost. Large companies have an edge there -- they pay less per wired-up employee for computer hardware, thanks to economies of scale. Big companies negotiate substantial discounts from PC vendors, and mainframes are actually cheaper per user than PCs. Similar economies apply to network hardware and to phone-line costs.
Too bad for large companies that hardware costs account for only about 15% of the total costs of running desktop computers. Software and administrative costs eat up the rest, and here the picture from a bigger company's point of view isn't pretty. Smaller businesses of almost any sort can pick up off-the-shelf, often industry-specific, software packages to handle accounting, sales, marketing, personnel, and distribution functions. Costs range from a few hundred dollars for a basic database or accounting package -- which can be more than enough for a tiny business -- to $50,000 or so for a highly sophisticated package aimed at midsize companies. Armed with basic programs, the average growing company can almost immediately start analyzing its customers, paring down inventory, producing slick marketing literature, and doing whatever else it takes to narrow the gap between it and its larger competitors.
But generic software for large companies simply doesn't exist at any price for many applications. Chrysler can't go out and pick up software for running a giant car manufacturer. And the software packages that are available for big companies, like accounting and personnel packages, often don't fill the bill anyway. That's because large companies typically have thousands of well-defined work processes that have evolved over decades and that involve tens of thousands of employees. The only alternative is to build application programs from scratch, at costs that can easily reach tens of millions of dollars. Even worse, the software needed to run a large company is so complex that it isn't uncommon for a company to abandon a program out of despair of ever debugging it -- after spending years and millions on it. BankAmerica and Pacific Gas and Electric are among the hundreds of organizations that have written off runaway software projects.
According to Jim Johnson of the Standish Group International, in Dennis, Mass., large businesses and government agencies spent about $81 billion on canceled software projects in 1995. And in 1994 Standish found that only 9% of large-company software projects are completed on time and on budget, and those that are completed have an average of 42% of the originally proposed features and functions. Small companies, on the other hand, complete 78% of their software projects, and those have at least 74% of their promised features and functions.
* * *
Who Can Adapt to Change Faster?
Then there's the task of updating software to meet business requirements that are changing ever faster. Smaller companies usually just throw out the old package and replace it with something state of the art, usually at considerably less cost than the previous version.
But rebuilding a huge homegrown system from scratch takes too long, so large companies try to patch up the old one, at tremendous cost and usually with little success. Forrester Research, in Cambridge, Mass., found that only 14% of Fortune 1,000 information-systems executives even try to claim that their core computer applications are flexible enough for business-process change. One simple example: now that many giant companies have invested millions in setting up Lotus Notes groupware on proprietary networks, smaller companies are finding they can achieve essentially the same capabilities for a fraction of the cost with programs that work over the Internet.
Some observers claim that large companies' software woes are quickly becoming a thing of the past as even the biggest organizations move away from hard-to-program mainframes to PC-based computing. But reports of the death of the mainframe have been greatly exaggerated. A study conducted by Sentry Market Research, in Westborough, Mass., shows that though the percentage of large-company computer-processing power coming from mainframes has dropped steadily from 80% in 1992, it still comprises 50%. Only one-third of large companies have as much as a single strategic PC-based application in place, and 40% haven't even begun the process of moving mainframe applications to PC-oriented systems. Large-company programming departments are so backed up with requests to fix or build systems that they're running two years behind -- essentially the same size backlog those departments were carrying six years ago.
