Virtual Focus GroupVirtual Focus Group

The world is littered with products nobody wanted. The Edsel. New Coke. Apple Computer's Newton.

Figuring out what customers want--what products with what features at what prices--has always been a challenge for product-development managers. Conventional methods of getting into consumers' heads, such as phone surveys and focus groups, are cumbersome, expensive, and time-consuming. And they're becoming more and more inadequate as time-to-market demands increase and product life cycles shrink.

In the fickle world of consumer products, companies are turning to the Internet as an alternative testing ground for new product ideas. Polaroid Corp. used the Internet to find out what features teen-agers wanted in the company's i-Zone sticker-photo instant camera. General Mills Inc. employs the Web to test ideas for new cereal and snack foods. Ford Motor Co. seeks consumer comments online about its plans to develop a sport utility vehicle with a gas-electric hybrid engine.

New-product market-research data traditionally is collected through surveys conducted by phone or through the mail, consumer interviews in malls, and focus groups. Using E-mail and Web sites, some companies are conducting surveys in half the time and at half the cost of phone or written surveys. Others are experimenting with more advanced techniques, such as Polaroid's use of Web-based conjoint analysis when designing the i-Zone.

Using the Internet for eliciting consumer suggestions on new products has advantages beyond speed and cost. Conventional survey results are consolidated by company marketers and then passed along to designers, often losing details or insights--and occasionally picking up biases--along the way.

The Internet can give design engineers direct, unfiltered feedback on new product ideas, says Nick Smither, director of IT for product development at Ford, which has been testing Web approaches to new product research since early last year.

The Web methods also can provide higher response rates and much richer responses. E-mail surveys often get response rates up to 20% better than phone surveys, which one product-development expert attributes to the fact that consumers can respond to E-mail surveys when it's most convenient for them. "The curt yes-or-no answers are replaced with paragraph-long responses," says Luis Ubinas, a principal at consulting firm McKinsey & Co. Another advantage of Web surveys is that respondents can better understand a new product concept through the graphically rich Internet.

But there are still some hurdles to overcome. There are worries that online survey participants don't truly represent the larger consumer population. And while some companies report cost savings associated with using the Internet, the relative immaturity of Web-based product development practices keeps costs up.

One effort to advance the use of the Internet for new-product research is the virtual customer initiative at the Massachusetts Institute of Technology's Sloan School of Management in Cambridge, Mass. Academics working with companies such as Polaroid and Ford are testing procedures to improve the speed, accuracy, and usability of customer input to the product-development process. Many of the methods have names that only an academic could love, such as "Web-based conjoint analysis" and "fast polyhedral adaptive conjoint estimation." But they're all refined means to answering the question of what new product features are valued by customers at what price. This issue of feature-price trade-offs is the most vexing problem faced by development and design engineers, says Ely Dahan, an assistant professor of management science and marketing at Sloan.

Conjoint analysis, for example, essentially asks consumers to design their dream product. A notebook computer maker can use conjoint analysis to find out whether consumers place more value on features such as speed, screen size, and disk space, or more value on cost or weight. Conjoint analysis isn't new. Market researchers have used the technique for years in face-to-face interviews with subjects.

But Dahan says that approach is slow and limits the number of feature combinations that can be presented to consumers. The Internet makes it possible to offer a greater number of feature combinations. Computer systems can even adapt to a test subject's preferences, eliminating from the exercise features that the subject clearly has no interest in. (MIT offers white papers and free demonstration software code at its Web site, mitsloan.mit.edu/vc).

Polaroid was an early user of this approach when it began designing the second generation of its successful i-Zone instant camera, which prints mini-photos that can be stuck anywhere. The camera is aimed at boys and, especially, girls, ages 13 to 18. Polaroid debated whether to add new features to the camera and increase its cost--the i-Zone retails for $17 to $25--or focus on its styling. Engineers at the Cambridge, Mass., camera manufacturer considered improving picture quality, auto-focus capability, and motorized picture ejection, a feature found on traditional Polaroid cameras that some technologists considered necessary. For style, Polaroid considered offering interchangeable front panels for the camera in funky colors and designs like those offered by some mobile phone makers.

To discover what teens really wanted in the i-Zone, Polaroid planned to use conventional market-research techniques, such as mailing questionnaires and showing drawings or mock-ups of the camera to teens gathered in shopping malls. But an intern from MIT working at Polaroid at the time proposed trying Web-based conjoint analysis. "Our market-research department wasn't really crazy about this," acknowledges Michael Rider, the company's director of technical operations for consumer imaging. One concern was that relying on adolescents with Internet access would provide a biased sampling of potential customers. The need to obtain parental permission was another hurdle.

Polaroid decided to experiment with Web-based conjoint analysis but supplement it with research based on conventional practices. The company created a conjoint-analysis Web site where teens were asked to assemble their ideal i-Zone. They viewed pictures of the camera showing how the proposed features would work and how it would appear with the stylistic changes. A counter at the bottom of the Web site displayed the increases and decreases in the camera's price based on the participants' actions. By making trade-offs between new features, styling, and price, the participants provided the camera maker with insight into what they valued and what they didn't.

Polaroid recruited several hundred teenagers, about 75% of them girls, for the experiment, primarily by using its main Web site and by recruiting teenagers at malls in five cities around the country. Polaroid also bought names from lists provided by a market-research firm. As an incentive to participate, each teen received a $5 gift certificate that could be used at online retailers. Polaroid provided the teens with passwords that gave them one-time access to the conjoint-analysis Web site. The collected data was loaded into a database using computer graphics interface scripts for analysis by Polaroid designers.

The results? Polaroid discovered that young teens viewed the i-Zone as a fashion accessory and the changeable face plates were most popular. "Customers wanted 'cool.' Clearly, style was important," Rider says. Conversely, teen-agers weren't very interested in the more advanced features that Polaroid engineers loved--including motorized picture ejection--if they added much to the camera's cost. Rider says test results obtained using traditional methods mirrored those from the Web-based conjoint-analysis experiment.

The updated i-Zone hit world markets in March and has been a hot seller--good news for financially struggling Polaroid, which sorely needed a winning product.

General Motors Corp. in Detroit is also working with MIT to take its product-development projects to the Web. This year, the car and truck manufacturer has twice created Web sites to study what it calls "basic vehicle architecture." Consumers, drawn from lists provided by third-party marketing firms, view architectural drawings and streaming video of new tailgate configurations for vans, ways of reaching the third seat in vans, and sun-roof and convertible designs for cars.

"Using the model, we can understand where customer needs aren't being met," says Andy Norton, a GM product research director. The survey findings, which Norton wouldn't disclose, will be incorporated into the design of future GM cars and vans. Until now, basic architecture tests involved creating vehicle mock-ups and bringing consumers to view them in large warehouses.

GM's Norton says the automaker paid consumers $50 to $75 to participate in its on-site basic vehicle architecture studies because the studies required several hours to complete. Because consumers can finish a Web survey in 30 to 45 minutes, GM believes those payments can go as low as $20--Norton says experiments offering $20, $30, and $40 have resulted in little difference in response rates. "A lot of people, given a good-quality survey, seem willing to do it for very little incentive," he says, contrasting the approach with dinnertime phone surveys.

Deploying the same software from MIT that Polaroid used with its i-Zone effort, GM began experimenting last month with a Web site it calls Truck Town, where consumers chosen from general marketing lists can design their "perfect" truck. Norton says the long-term goal is to make the site open to all consumers.

Some companies use the Net to test new or revised products. ReplayTV Inc., which manufactures a device for storing TV programs on a hard drive, used consumer-testing services from BetaSphere Inc. when beta testing the second generation of its product early last year.

BetaSphere maintains a list of thousands of consumers it recruits through its Web site to voluntarily test new products. It offers the occasional free T-shirt or product discount, but most people are drawn to the site to see the latest new products. ReplayTV in Mountain View, Calif., turned to BetaSphere because it lacked the time and in-house expertise to create and manage a beta test program on its own, says Kimberly Wiefling, a consultant who was managing ReplayTV's product development efforts in 1999 and 2000.

ReplayTV sent out beta versions of its improved system to 120 consumers. Half the consumers participating in the beta test had some technical expertise; the other half were more mainstream consumers. BetaSphere built a Web page for each consumer through which they were given weekly "assignments" such as installing the product and trying out its slow-motion replay capability. The consumers then recorded their comments on their Web pages, including what they liked about the product, what enhancements they would like to see added, or what bugs they found.

The testers' discoveries included a software problem that caused the system to freeze up when its memory was full, a bug that the company corrected before the final product shipped. Other user comments, such as the need to improve the ease-of-use of ReplayTV's remote control, are being incorporated into future versions of the product. The whole project cost about $50,000, Wiefling says.

These experiments represent the leading edge of using the Internet for customer research. Companies that aren't as far along are still speeding up research efforts by conducting consumer surveys by E-mail and setting up Web pages to solicit new product ideas.

Soft-drink maker Coca-Cola Co. in Atlanta, for example, uses Ideas.com, an Internet site for soliciting ideas from consumers, as part of its effort to expand its sales beyond carbonated soft drinks. Last September, Coca-Cola posted a notice on Ideas.com seeking ideas on how to get children to gulp more sports drinks or fruit juice. The winning idea--which earned a $5,000 prize--was to develop a wrist-wearable package for toting the beverages. Coca-Cola is examining the feasibility of the concept.

Coca-Cola's goal is to bring consumers into the product-development process from the start, rather than develop product concepts internally and then test them on consumers. "If you've got a good idea, you've got a good start toward a successful product," says director of innovation Tom LaForge. Another benefit is that the Web site brings ideas from all across the country, rather than just the six or eight cities that would be surveyed using conventional market-research methods. Those who respond tend to be more passionate than test subjects randomly selected for interviews or focus groups, LaForge says.

Ideas.com collects consumer responses from its Web site and loads them into a searchable database that it provides to its clients. Earlier this month, Coca-Cola used the site to solicit product formulation, packaging, and distribution ideas to get customers to drink more of its Dasani bottled water. Sears, Roebuck & Co. is also on the site, looking for ideas on new yard tools, as is International Paper Co., searching for novel packaging concepts.

Ford is making extensive use of the Web to get customers' voices into its product-development process. Some sites are open to anyone, such as http://www.hybridford.com, where consumers can comment on Ford's plans to develop a version of its Escape SUV with hybrid engine technology by 2003. Earlier this month, for example, the site had a spirited discussion about whether a diesel version of the vehicle would be better or worse for the environment. Other sites are restricted by passwords, such as those for Focus and F-Series pickup owners, who comment on what they like or don't like about their new vehicles. The automaker's marketing organization and engineering teams review the comments.

General Mills used to rely on phone and mail surveys and "mall intercept"--questioning people in malls--techniques to query consumers about existing products and test new-product concepts, such as new flavors of its Yoplait yogurt. But the Minneapolis breakfast-foods maker now uses E-mail for 60% of its consumer survey work. And, of course, the Internet won't eliminate the need for live consumer taste-tests once a new product is developed.

Speed remains the biggest selling point for Internet-based market research. "We've literally turned around studies in a day or two that would have taken a month or more using traditional approaches," says Sloan's Dahan.

Using the Net in the early stages of product development also holds the promise of cutting market-research expenses, such as employee travel to test-market cities and the costs of hiring phone-survey companies. Gayle Fuguitt, VP of consumer insights at General Mills, says the cereal maker has cut its consumer survey costs by 50% by using E-mail.

"Today, companies pay thousands of dollars to experts to come up with new product ideas," says Philip Tulimieri, a director in Deloitte & Touche's technology and communications practice and a market-research expert. "With the Internet, companies can set up a site asking customers what they would like to see in new products."

Despite the i-Zone success, Polaroid has stuck with traditional product-research techniques. Rider says company marketers remain concerned that Web surfers might not be representative of the general population. Rider also notes that while the demographics of those selected to participate in Web tests can be controlled, the demographics of those who ultimately log on to participate cannot.

Costs have remained an issue for Polaroid. The i-Zone experiment added several thousand dollars to the development project's total cost because the company used conventional market-research practices in addition to Web-based conjoint analysis for verification purposes. Rider says that using the Internet-based research won't be cheaper in the short term because it requires a bigger pool of test subjects than traditional methods.

Security remains another hurdle for Web-based product-development methodologies. GM doesn't provide actual product designs or sketches on its basic vehicle architecture or Truck Town sites. Norton contends that providing such details to selected consumers through a secure Web site isn't much riskier than bringing them into a GM site and showing them pictures. For the immediate future, though, he expects "a lot of organizational resistance" to the idea.

But Polaroid's Rider says that as these problems are worked out and since Internet use is becoming more prevalent, using Web-based consumer-survey practices to get the voice of the customer into the product-development process will become more widespread. "We saw this as a worthy peek into the world of online research," he says. "This is a young idea, and it's still coming along the learning curve."

Nevertheless, these practices are steadily becoming more mainstream. Serious use of the Internet for product-development applications has only been going on for about 18 months. "They're still in their infancy," Tulimieri says, "but there's enough evidence out there that indicates these practices will work."