(FORTUNE Magazine) – KNOW thy customer! Unfortunately, heeding that rule of commerce is no easier than obeying the classic rule of life: Know thyself. Efforts to get to know the consumer cost a bundle, but many just don't seem to work. An executive commented in a poll by the Center for Strategy Research in Cambridge, Massachusetts: ''Market research can be extremely valuable, but most of what is done right now is so poor as probably to be fraudulent.'' And expensive. Last year companies handed over a record $3.5 billion to the 50 biggest research firms to learn more about their customers, says the Honomichl Report, the research industry yardstick. Measured by the luck companies have had introducing new products, they wasted a bundle. Group EFO, a research outfit in Weston, Connecticut, figures that in 1992, 83% of new products failed to clear the performance hurdles set by their developers. Raising the batting average begins with the basics. Rule No. 1: Don't overlook the most obvious way to get to know your customers -- talk to them. Companies are exploring new ways of doing this. Rule No. 2: Mine the data about your customers that are already in your computers. Rule No. 3: Don't let focus group results overwhelm your common sense. Honda uses all the standard research tools, including focus groups and customer surveys. The company even videotapes drivers as they test new cars. In response to all this customer input, Honda has made thousands of changes in the Accord since introducing it in 1976. The new wrinkles range from installing a suspension system used on racing cars for improved handling to changing the shape of the rear window so that a large soft drink can be passed into the car without spilling. In the process Honda just happened to produce the best-selling car in America from 1989 until 1992, when the Ford Taurus edged it out. Says Ben Knight, vice president for R&D of Honda North America: ''We believe that the market and the customer will always find the truth.'' Honda's manufacturing unit kicked off its most extensive customer research effort yet, the E.T. Phone Home Project, in November 1992. (The name and the notion were lifted from the film E.T.) Over the following three months, factory workers who actually bolt and bang the Accord together called over 47,000 recent Accord buyers, or about half the owners who registered their cars with the company the previous spring. Honda's goal: to find out if customers were happy with their autos and to get ideas for improvements. Given the long lead times in the auto industry, you can expect to see the impact of E.T. on the 1995 and 1996 Accords. Old-fashioned surveys done in a Texas-size way keep USAA, the San Antonio financial services company, in touch with its 2.5 million customers. The company, which markets insurance, banking services, and mutual funds mainly to military officers and their families, mails multiple-choice questionnaires to 500,000 of them every year, inquiring about customer satisfaction, future needs, and ideas for new products. In a world where a 20% response rate is pretty good, 60% of those polled return completed questionnaires. More important, USAA follows up. Based on what consumers said they wanted, the company launched a growth and income mutual fund last June. It was an instant success, attracting $76.8 million in assets in less than five months. Now USAA is moving away from the multiple-choice format to more open-ended questions. So are many other companies. Thomas Dupont, president of D 2 Research in Mountain Lakes, New Jersey, says open-ended mail and phone surveys can cost two to three times as much to conduct and process as the close-ended type, but they are pound-wise. When a company writes all the allowable answers, explains Phillip J. Riese, president of the cardmember financial services group at American Express, ''the survey can become a self-fulfilling prophecy. Open-ended questions allow customers to tell you something that is not on your list.''

HEWLETT-PACKARD recognizes that customers volunteer a tremendous amount of information when they contact the company to file a complaint, make a suggestion, or request help. So every piece of customer feedback, corporate and individual, is assigned to an ''owner'' at H-P who must act on the information and report back to the person who called. If, for instance, a customer beefs about an H-P printer, someone in that division will be assigned to check the company's database to see whether the complaint is widespread and, if so, what the company is doing about it. The ''owner'' then gets back to the customer. Out of this process came the PerfView networking software. H-P was selling software that monitored the processing speed of its networks. Customers reported that the product worked as advertised but complained that it didn't do more. Now PerfView has been redesigned to keep track of the workings of the network's individual components even if the machines are manufactured by different companies. Tucked away in the bowels of many a computerized billing system are megabytes of data about what customers order, in what size, and in what quantity. This reveals not just what people say they want, but also what they are willing to pay for. Spiegel, the big Chicago-based direct-mail retailer, mined its database and hit gold last summer: E Style, a catalogue aimed at black women. To create it, Spiegel hawked its regular wish book in Ebony and Essence magazines to create a mailing list. Then, using its database, Spiegel kept track of these customers' purchasing patterns, using what they bought as a guide to putting together a catalogue with specific appeal to African American women. E Style, for instance, offers a lot of hats because Spiegel's database shows that black women buy them. Marketing vice president Karl Steigerwald reports that E Style sales are running 50% above the company's original projections. Powerful desktop computers and user-friendly software let any business do bigtime database marketing, such as customized mailings based on previous purchases. Pizza Hut outlets use customer lists, derived from their takeout and delivery business, to target customers with promotions tied to consumers' tastes. Pepperoni fans, for instance, may get a coupon in the mail for a special deal on their favorite topping. As part of processing a purchase at Radio Shack stores nationwide, clerks ask customers for their names and addresses and enter the information into PCs right on the sales counters. Later, customers receive mailings tied to their interests. Marketers have always known that subconscious forces drive us to consume, but understanding the deep aspects of human motivation is difficult. Focus groups, which have been around since the end of World War II, can make the job a little easier. But new psychological techniques are competing with them, and new academic thinking indicates they may not be as cost-effective as more traditional qualitative research. Given the way that focus groups work, it is easy to see why Madison Avenue likes them. Market researchers gather ten or so customers to talk about a product or service, usually without telling them who the sponsor is. While a moderator leads the discussion and elicits the group's opinions, good or bad, about the product, researchers, and usually some employees from the client company, observe the proceedings from behind a one- way window. The consumers' comments are taped, sometimes videotaped, and transcribed. Researchers then comb through the transcripts looking for common themes. Those transcripts also provide rich fodder for ad copy. Companies can best use the studies to make sure they are not alienating a customer. For example, a paper-products company killed a toilet tissue commercial it was planning to run nationally when a focus group in the South perceived sexual innuendo in the central image -- a 7-year-old girl in a nightgown on a bed -- that the ad agency didn't see.

BUT PACK some skepticism with the takeout food you bring behind the one-way window when you observe your customers. The sample of opinions in a focus group, far too small to be statistically representative, can shrink further owing to the Twelve Angry Men effect. Someone in the group ends up acting like Henry Fonda in the classic movie, persuading the other jurors to change their minds. Focus group research should not be used for making decisions without further validation from other methods of research. Jeffrey Jury should know. He directs marketing for InterMountain Canola, a vegetable-oil maker in Cinnaminson, New Jersey. Focus group research he had done for potato chips made with InterMountain's canola oil, a relatively healthful type, said the advertising should stress how good the chips were for you. But Jury was not persuaded. A healthful potato chip is nearly as oxymoronic as a smokeless cigarette. To check his suspicions, Jury brought in Stan Gross Associates, a research firm from Haverford, Pennsylvania. It tries to get into what President Stan Gross calls customers' inner minds by having a group of participants play games -- literally -- like Nerf basketball and blowing bubbles, which loosens them up so they express themselves more revealingly. Rather than simply asking for consumers' opinions, Gross tries to divine their feelings and attitudes by, say, asking a group to write stories about products. His research showed that what customers liked most about canola oil was that it could make them feel less guilty when they pigged out on junk food, a sentiment they were reluctant to voice directly. InterMountain changed the language used to promote the product from ''Choose to be healthy'' to ''A taste you can feel good about.'' And the chips were a hit. Not only must focus group research be buttressed with other types of information, but it is also beginning to look as if it is no better than old- fashioned, low-tech, ''ask the man who owns one'' consumer research. Marketers usually assume that focus groups are more efficient, and therefore cheaper, than lengthy individual interviews because people interact in a focus group, bringing more needs to the surface quicker. That helps keep costs down. Not so, says a recent study by Abbie Griffin of the University of Chicago and John Hauser of MIT. They compared the customer information about a complex piece of office equipment drawn from eight two-hour focus groups and nine one- hour one-on-one interviews. They found no difference between the two methods in the amount of information produced or the speed with which it surfaced. Confirming Griffin's findings will no doubt require more work. Should they turn out to be true, however, perhaps the IRS could be convinced that the ''research'' that you have been doing with customers over lunch at your favorite expense-account bistro ought to be 100% tax deductible.

BOX: LOVE IS NOT ENOUGH

Though boomers adored the first Accord (above) in 1976, Honda kept talking to customers about improvements. All the tweaks add up in the new 1994 model.

1. The engine size nearly doubled, to 145 hp. 2. Wheelbase grew 13% for better rides. 3. One key operates all four door power locks. 4. Racing-car suspension for improved handling 5. A sportier look as the front hood noses down 6. A holder for the garage door opener