An increasingly popular research technique helps marketers and consumers get what they really want.
This past November, the Lands' End Web site launched "My Personal Shopper," a recommendation engine for customers who want help sorting through the retailer's vast selection of sweaters, skirts, and button-downs. Big whoop, you say - Amazon's been doing this for years. But unlike companies that use past purchases to proffer suggestions to cyber-browsers, Lands' End is the first apparel retailer to use a technique called conjoint analysis. In a brief survey, six pairs of outfits are shown to the shopper, who chooses a preferred outfit among each pair. Through analysis of these six simple choices, and the answers to a few other questions, the site sorts through 80,000 apparel options and presents the most suitable ones to the busy shopper.
While the use of conjoint analysis by Lands' End is unique, the methodology itself is not. It's a research technique that has been around for three decades, but which is increasing in popularity as software developments and the Internet make it easier to use, as well as more powerful and flexible. Understanding how conjoint analysis works, and the innovative ways it's now being used, provides a good opportunity for any company to increase its chances of giving consumers more of what they want, and less of what they don't. "Use of this method will increase as more marketers realize what it can do, and how well it can work," says John Seal, senior analytical consultant at Burke, Inc., a Cincinnati-based research firm.
So what is conjoint analysis? The rationale underlying the technique is that consumers weigh all the many elements of a product or service - such as price, ingredients, packaging, technical specifications, and on and on - when choosing, say, a sweater, airline ticket, or stereo system. While this may seem obvious to anyone who's faced a wall of DVD players at Circuit City, figuring out how to leverage this concept in the marketing arena can be difficult. Conjoint analysis does this by breaking products down into their many elements, uncovering which ones drive consumer decisions and which combination will be most successful. But rather than directly asking survey respondents to state the importance of a certain component a la traditional surveys, participants judge hypothetical product profiles, consisting of a range of defining characteristics called "elements." Their responses are run through an analytical process that indirectly identifies the importance and appeal of each element, based upon their pattern of preferences for the element groups.
If this process sounds more complicated than a traditional survey, it is. And it tends to be more expensive as well. But, as the saying goes, you get what you pay for. While traditional surveys can gauge interest in product features, the results can be misleading. This is because it can be difficult for respondents to directly relate how valuable a particular product feature will be to them. "If you ask respondents how much they are willing to pay for a certain feature, they often can't or won't answer truthfully," says Tom Pilon, a Carrollton, Texas-based consultant who specializes in conjoint research projects. "They'll tend to say they're interested in all the new features." They wouldn't be lying, but they might not actually pay for those features when the product comes to the market. Similarly, focus groups are a good way to draw out consumer opinion on new products, but it's difficult to accurately quantify how a product will perform in the marketplace from this data.
"Conjoint mimics the way that consumers actually think," says Joel Greene, director of database marketing at Akron, Ohio-based Sterling Jewelers. Greene first used conjoint research last spring, and is impressed with the results. Fed up with consumers tossing his mailings into the trash, Greene hired White Plains, New York-based market research firm Moscowitz Jacobs Inc. (MJI) to figure out a way to make them more appealing. Using a proprietary research tool called IdeaMap, MJI worked with Greene to systematically break down the brand image and communication efforts of Shaw's (a division of Sterling Jewelers) into bite-size elements. These factors were culled through focus groups and brainstorming sessions that examined previous marketing efforts and possible new approaches. Well over a hundred elements were part of the tested pool, which included different ways to convey messages about Shaw's stores, merchandise, brand differentiation, and emotional appeals. "We wanted to cast a wide net, because we didn't know what would work," says Greene.
MJI recruited a group of more than a hundred survey respondents to its testing facilities in Chicago and White Plains. Seated at computers, they were systematically exposed to the different elements, grouped as words, phrases, and pictures. For each random grouping of elements, the respondent would rate the appeal of the group as a whole. From an analysis of the pattern of ratings, MJI was able to give a utility score to each element. Using these scores, Shaw's could then create marketing messages from this universe of elements appealing to the widest group of customers, or to specific segments. The words, phrases, and pictures (i.e. elements) that scored highest for each segment were then used to create new mailings. And the glittering result? The creative geared toward each segment resulted in significantly higher rates of response, as well as increased dollar sales per response.
The effectiveness with which conjoint can be used to understand precisely which aspects and features of a product are driving sales is especially crucial in an industry such as consumer electronics. With an increase in digital convergence, and with hybrid electronic products coming to the market - think refrigerators connected to the Internet, and cameras as MP3 players - the question arises: Will consumers actually pay for these products, and how much? "We really have to avoid the `if you build it, they will come' pitfall," says Maria Townsend-Metz, a marketing manager at Motorola.
Heeding this warning, Townsend-Metz used conjoint analysis while working on enhancing Motorola's popular TalkAbout two-way radios. "We couldn't put all the different options we were thinking about on the radio, so we needed to know which ones were going to be of most value to the consumer, and help sell the most radios," says Townsend-Metz. Because of the complexity of creating and modeling well-run conjoint studies, she brought in Boise, Idaho-based research firm POPULUS, Inc. In six markets across the U.S., the company conducted conjoint surveys of consumers who participated in activities, such as camping and biking, where a two-way radio would be a natural accessory. POPULUS tested 18 attributes, covering technical specifications, price points, and the appearance of the devices.
Using a conjoint methodology was especially appropriate because all the attributes were interdependent - different features, for example, would affect the look of the radio, as well as the price. "The goal was to find the combination of features that would maximize interest at the lowest production cost," says John Fiedler of POPULUS. The resulting product was right on consumers' wavelength, and the TalkAbout now leads the market for recreational and industrial two-way radios.
The popularity of conjoint research was greatly increased by the development of software in the 1980s that made it easier to design and run these types of studies. The leader in this field is Sequim, Washington-based Sawtooth Software, whose ACA brand of conjoint is the most widely used in the world. Other software suppliers include SPSS Inc. and SAS Systems. Prior to computer-assisted research, conjoint surveys were conducted using cards that had groups of attributes printed on them, and which were sorted by preference. The number of attributes that could be tested in this manner was severely limited, as was the concluding analysis.
The trend toward conducting survey research on the Web will further increase the use of conjoint, according to experts in the field. The Web provides an easy way to present respondents with groups of attributes, something that was much more difficult to do over the phone (people can only remember so many features at once). Fuji Film, for one, has used conjoint Web surveys to uncover the effects of price, brand, and package configurations (i.e. the number of rolls in a package) on sales. "Film is a low-involvement category, the product is standardized, and the effects of price and packaging are significant," says Doug Rose, president of Austin, Texas-based DRC Group, who worked on conjoint projects last year for Fuji.
By showing respondents side-by-side attribute profiles of different brand, price, and packaging configurations, Fuji was able to analyze their patterns of preference, and deduce what was driving their choices. The film manufacturer was further able to estimate exactly what effect a certain price point on a particular package of film would have on market share. This conjoint study was so accurate that its estimates perfectly matched ACNielsen data on price elasticity in the film sector, which appeared after the Fuji study.
One research firm taking conjoint analysis a step further on the Web is Burlingame, California-based Active Research. Its proprietary "Active Buyer's Guide" is a powerful research tool for marketers, disguised as a shopping search engine for consumers. Licensed to over 70 popular sites, such as Lycos and MySimon, it helps Web shoppers find the computers, appliances, and financial services (135 categories in all) that most closely match their needs, both online and offline. By filling out a conjoint survey that hones in on what features, price points, and attributes they are looking for, the Guide delivers a list of products that are most likely to interest the shopper.
But Active Research doesn't do this just to help out consumers. By answering the questions required by the search engine, shoppers are providing the company with a gold mine of continuous information on what kind of products they want, and at what price. In effect, Active Research is compiling 1.5 million surveys a month. What's more, these surveys are from people who are providing the most accurate information possible and are in the market, at that moment, to buy a particular product. By compiling and analyzing this data, Active Research provides up-to-the-minute information for clients such as Ford, GE, and Sony on which aspects of a product are driving consumer decisions, which demographic segments are driving sales, and who's interested in different features.
In addition, clients of Active Research can create hypothetical products and measure what their likely market share would be. Using the conjoint-produced utility scores of different product features, marketers can preview how a new product will sell in the marketplace, without the time and cost of a test launch. Because of the size of its sample, Active Research can slice-and-dice hypothetical products in an array of categories, demographics, and configurations. "It's not an exaggeration to say that what they are doing is an absolutely unique way to do primary research," says client Suzanne Snygg, futures product manager at Palm, Inc. The dual nature of their service is highlighted by the fact that Snygg herself has used Active Research data not only to shape product concepts for Palm but also to find the best mini-stereo system for her home. As the Web makes conjoint analysis more popular, it's important to note that conjoint research is still more complicated to conduct than straightforward survey research. To produce worthwhile results, it is crucial to create a pool of attributes that actually influences consumer choice. This requires careful and creative brainstorming. Re-searchers have to choose the correct conjoint method (there are several types, with many researchers creating their own unique variants). They have to show groupings of elements to respondents that cover many possible combinations, in a balanced and useful way. The final results are only as good as the design and analysis of the research, which can be complicated. Keith Chrzan, director of marketing sciences at Maritz Marketing Research, goes so far as to say that "a lot of people are using conjoint who shouldn't be," due to the easy-to-use software.
That said, the effectiveness and accuracy of conjoint techniques make them powerful tools for marketers who use them properly. Says Tom Pilon, the Texas-based consultant, "once a company has done it once, they always come back for more."