SOFTWARE EVEN A CFO COULD LOVE
The computer, that brilliant fool of modern times, is gaining a measure of street savvy--and that's sparking fresh interest among high-level corporate finance types. Chief financial officers long relegated the computer to the back office, where it mindlessly crunched numbers for routine matters such as customer billing. But now it can function as a strategic planning tool and all-round financial adviser, making it suddenly welcome in the corporate suite. There is new software available that permits the computer to mimic human reasoning--and in a user-friendly fashion acceptable to even the most hardened Luddite in the front office.
CFOs are increasingly benefiting from the computer's ability to offer predictions, spot trends and trouble spots, and present options. Want to know the cheapest way to ship goods--this week and next June? Or where the waste is that keeps cash flow subpar? Or what capital spending makes sense strategically? Or what's the best employee stock-ownership plan?
`GREAT SHORTCUT.' While only a handful of companies are currently taking advantage of the software, the list is growing. "You're seeing this intelligence beginning to creep into all kinds of applications in the financial area," says Charles A. Christy, senior consultant with Arthur D. Little Inc., a Boston consulting firm. Most users are large corporations that need to track vast revenue flows and international currency fluctuations, such as AT&T, DuPont, Lockheed, Sears, and Xerox. Other, smaller users, such as Mrs. Fields Cookies Inc., find getting into this realm easy because it's not expensive. The cost to set up a system at a small outfit is about $200,000.
True, a machine isn't about to replace the CFO. And no one is suggesting that a cybernetic brain will be making any final decisions for people. The day when a computer will actually think, like the nefarious Skynet in the Terminator movies, is far off. Today's "artificial intelligence" has several more modest variants that differ by degree of sophistication (table). Each promises quick and accurate sifting of data to find patterns that CFOs can act on, saving them time and headaches. "This is a great shortcut," says Randall K. Fields, until recently Mrs. Fields's CFO and now the cookie retailer's computer honcho.
Building bona fide wisdom into an electronic box is quite an advance, first commercialized in the early 1980s with methods developed at Stanford, Carnegie-Mellon, and Massachusetts Institute of Technology. Until then, the only strategic role for computers was rapidly calculating numbers, retrieving information, and sorting and ranking data. Although that spadework was helpful--an army of clerks would need weeks to do it--executives still had to figure out what all the facts meant and how to use them. Says Gerald D. McInvale, CFO of New Orleans-based Entergy Corp., an electric-utility holding company: "One of the blessings of the Information Age is that you have a lot of data. One of the curses is that you have a lot of data."
The new software gives the computer judgment. In Corporate America, the most popular and user-friendly artificial-intelligence technique is the "expert system," which injects the experience of seasoned experts--rules of thumb and tricks of the trade as well as knowledge--into the soulless circuits. This technology winnows out irrelevant factors and arrives at practical alternatives. Says Patrick Perez, chairman of software maker Neuron Data Inc. in Palo Alto, Calif.: "Without expert-system help, you cannot filter out the complexity." Using the past as a guide, such programs can deal with the avalanche of minor decisions that executives make daily. More advanced systems, such as neural networks, can learn by themselves from their mistakes. When the computer projects a 10% profit gain and the results come in at 5% because of an unforeseen slump in product demand, the computer learns from that and remembers next time.
These smart programs behave like a patient and polite aide. "They don't respond with: 'Here's the answer, and the answer's no,' " says Arthur D. Little's Christy. The computer, he says, "will tell you what it saw, suggest how to solve a problem, and point out the reasons it is suggesting this."
BROKEN PROMISES. Artificial intelligence, however, has suffered a rocky beginning during its almost 10 years of commercial availability. Many of the early AI software houses went out of business, about 20 of them around Boston alone. In fact, the term "artificial intelligence"--an umbrella expression for expert systems, neural nets, and other techniques--has fallen into disrepute and is seldom heard anymore. Early users found that the software was overly expensive and, worse, often promised to deliver far more than it really did.
What's more, the technology scared a lot of CFOs, who are accustomed to having information brought to them, not calling it up themselves--a problem that persists, despite more accessible formats. Says Nancy Martin, a Coopers & Lybrand partner: "Walk into the office of a typical CFO, and you won't see a computer anywhere." This has a political dimension, too: The CFO, who usually prides himself on knowing everything that's going on, has a tough time working with the computer jocks who oversee the system. They "sit in his office, using terms that he doesn't understand, and make him feel stupid," says one corporate financial official.
The systems won their first converts among financial companies--banks, insurers, brokerages--which must grapple with mammoth transaction volumes. They focus more on the frontline operations than high-level strategizing. They are employed by banks to evaluate loans, insurance companies to underwrite policies, and brokerages to plan clients' portfolios and trade issues.
American Express Co., for instance, has installed expert systems to help its legions of credit authorizers, who sit at computer terminals and decide whether to O.K. a charge. That's a crucial function at AmEx, which has no credit limits on its cards. Before the new technology kicked in, an authorizer would look at as many as 16 screens of data per customer. The company felt this was cumbersome and that too many bad calls were made. So AmEx distilled the knowledge of its best authorizers into a program that speeds up and sharpens the process.
SIMPLE SOFTWARE. While nonfinancial companies are taking longer to appreciate the virtues of AI, they are coming around. A big obstacle has been that programmers had a rough time writing good software for corporate finance. Reason: Different industries and even individual companies have unique circumstances. A bank loan applicant can be judged by a fixed set of variables; a corporate income forecast, on the other hand, has far more unknowns to contend with, such as the strength of rivals' marketing drives.
Lately, though, a myriad of software is on the market to handle the entire spectrum of business. Digital Equipment Corp., for example, has a special system that follows the 23 steps of selling a machine, from accepting an order to receiving customer payment. Before AI, DEC had to wait a while between shipping the product and recognizing the revenue. Late payers and nonpayers always made booking revenue tricky. These days, DEC knows in a flash what its actual likely receipts will be.
The growing power and user-friendliness of PCs nowadays is allowing the technologically benighted to partake of their wonders. Software is much less intimidating and, unlike early AI, doesn't flaunt its expertise. Programs now can walk executives through many complex operations that formerly required all sorts of abstruse commands. "The program does most of the work," says David Fein, vice-president of San Diego's ValuSource Inc. software company. The labor-saving aspect of computers shines brighter in the present era of corporate downsizing, when fewer people are on hand in headquarters to analyze data. Says Jim Garnard, CEO of Trinsic Corp., a software producer: "They have to do more with less."
Changes in corporate structure have made AI more acceptable as well. The steadily growing presence of computers has forced many companies to name chief information officers, who often are on the same level as CFOs. This brings computers "out from under a very conservative drudge," like the often tech-loathing CFO, says Arthur D. Little's Christy. Plus, computerized consulting firms have arisen to help out corporate clients, such as Dallas' Cost Containment Marketing Group Inc., which assays values of closely held companies up for sale. ValuSource software, says Cost Containment CFO Doug E. Childress, helps churn out a report on a company by "automatically pulling out numbers and then throwing them into the text."
Companies that do the job themselves find that the saving in time and money is the most immediate payoff. "You can't wait two weeks to say we'll figure out how to do something," says Elizabeth Jennings, CFO of Kerr Steamship Co. in Cranford, N.J. "If we didn't have the system, we would have to triple our staff." One way Kerr uses expert systems is to discover the cheapest means of transporting inland cargo. Rates and routes are built into the computer, which calculates the best itinerary, much like a telephone network with automatic routing. Rail rates get updated daily. Before, Kerr staffers had to spend huge amounts of time hunched over manuals, calculators in hand.
Freed of this deadening scut work, managers can make the best use of their minds, deciding at a glance what to do next. District managers for Mrs. Fields used to take 4 1/2 hours weekly to gather all the numbers from a single cookie outlet. With 10 stores in a typical district, that meant 40 hours each week. Now, the company's central office gathers all the data, and district managers can spend more time at the stores. "The highest and best use of technology is to get people dealing with people," says Randall Fields. "We're trying to get all the paper functions obliterated."
FARSIGHTED. The expert system helps Park City (Utah)-based Mrs. Fields cut down on errors. With 800 stores, the company has a real challenge keeping track of its kaleidoscope of ever changing rental expenses. Not only are leases expiring all the time, but outlets further complicate the picture by renting extra storage space nearby. The potential for bad data could drive an accountant mad if the software weren't there as a backstop. When someone enters the wrong number for a store, the computer will spot the mistake. "The expert system can check each lease," says Fields.
Computers make financial planning much more flexible than before. Take Xerox Corp. Until two years ago, the copier giant's seven-member sales-forecasting team used a hodgepodge of methods to do their work, ranging from pencil and paper to computer spreadsheets. They started on the 1991 forecast in mid-1990. Now, they're using an expert system, which gave them the luxury of waiting until October to start on the 1993 plan. "We're trading information electronically, instead of passing around pieces of graph paper," says Thomas F. Applebee, a forecast team member.
The new system lets Xerox project three years out; previously, the forecast team could only manage the next 12 months. And it constantly analyzes actual results against the forecast, updating it. Further, the time saved allows the Xerox team to consider how factors outside their own company might affect it, such as inflation or rivals' performances. Elsewhere in the company, expert systems give advice on marketing: how new products move vs. older ones, how price reductions will impact sales, how compensation changes can better motivate salespeople.
Writing these programs is far from easy. Xerox has spent the past year working on one aimed at giving local sales managers a profit analysis for each deal. Since the deals can be as individualized as snowflakes--depending on service contracts, purchase financings, supplies, and the like--a manager in the field might not know at once just how much he is reaping on one.
`LESS FUSS.' Widespread computerization cuts in two directions. The head office knows much faster what's happening in the field. But the field also can quickly find out the hot skinny companywide. Mrs. Fields's Fields muses that this has cost top executives some power. As CFO, he says, time was when he could tell a subordinate in another state: "I know things about the business that you couldn't possibly know." Nevertheless, Fields applauds the power shift because the rank and file now can respond in an instant. If a problem crops up, he says, "they're probably in motion about the same time as I'm finding out." Freedom from daily minutiae, he adds, could change the role of the CFO, making it less tactical and more strategic.
Although the computer may empower subordinates, that doesn't mean that they are free to run amok. Lockheed Corp. has a program that ensures that company procurement procedures are followed to the letter. Scientists at Lockheed's research and development facility in Palo Alto don't have to submit forms through the bureaucracy to order material for their experiments; they can do this right on their computer screens. But the program has a safeguard: It assists the scientists in filling out orders, walking them through all the different coding so they don't charge, say, industrial coolant to the wrong account. Good thing. "We've got lots of projects going at any one time," says Harold D. Leister, head of material services at the Palo Alto facility.
Computer enthusiasts see limitless possibilities for AI. Entergy employs computers to monitor cash flow and oversee travel and expense accounts. Next, the utility concern's McInvale says, Entergy will install programs for maintenance scheduling on its nuclear plants and for authorizing capital spending. Ending the mind-numbing complexity that goes with those tasks, he says, will "let employees do their jobs better with a little less fuss." Artificial intelligence, built in the image of its creator, is a worthy servant indeed.A NEW AGE GLOSSARY
Software systems that attempt to replicate aspects of human intelligence. Term
has fallen into disuse because early AI systems did not live up to their promise
A computerized decision-making technique that embodies knowledge gleaned from
experts. It helps users sort out a blur of factors and choose the right option.
Examples: forecasting, valuation
A more sophisticated system that tries to mimic human brain processes and
learns from mistakes it makes. Used for predicting such things as market price
movement and lending risk
Holds that seemingly random events, such as stock prices, actually have
patterns that canny computer programs can detect. Used to analyze the behavior
A problem-solving technique useful in identifying and handling anomalies.
Bowing to Charles Darwin, outcomes judged unfit do not survive to remain in
Larry Light, with Chris Welles, Suzanne Woolley, and Leah Nathans Spiro in New York