Editor’s note: William H. Ducker is senior partner, Ducker Research Company, Inc., Bloomfield Hills, Mich.
Conjoint analysis, a sophisticated form of trade-off analysis, is widely used in consumer marketing research. It also is a valuable tool in industrial and business-to-business marketing research. We have found conjoint analysis particularly effective in studies involving product development strategies that are quite technical in nature. We have also found the technique a useful supplement to analyses of the purchase decision process in industrial markets.
In nearly every instance, there are far too many variables to permit manual filling out of a questionnaire. It is not unusual to have 50,000 possible combinations of product features and attributes in a product development study or 50,000 combinations of purchase selection criteria. In these studies, an adaptive conjoint analysis program (ACA) is essential.
The ACA system is a microcomputer-based interviewing procedure which was designed for conjoint studies that have a large number of attributes. The interview is customized for each respondent, focusing on the respondent’s most important attributes and most relevant attribute characteristics. As the respondent proceeds through the program (about 10 minutes), questions become more complex; intermediate questions and closing questions on buying intentions are programmed to reflect earlier replies from that particular respondent . . . hence, the term adaptive. (There are several ACA programs on the market. The reader’s local computer service outlet should be able to help. You may also wish to consult the marketing research software directory in the February 1996 issue of Quirk’s Marketing Research Review.)
Pilot study
Prior to a full-scale study launch, a pilot study is conducted. In these pilot interviews, respondents are asked to define attributes they consider when specifying or purchasing the product under study. It is particularly important to define levels within each attribute, such as horsepower ranges, for example. These attributes and their several levels must be practical and actionable; it means nothing if "wish list" items are included. The ACA software then is customized to reflect results of the pilot interviews. The complete interview is on diskette and is thoroughly tested prior to beginning fieldwork.
Respondents asked to participate in ACA need not be computer literate. All they need is access to a PC and the ability to turn it on. The following are examples of recent conjoint studies we have conducted, listed alphabetically: auto components, controls, doors, flooring, insulation, lighting, roofing, valves, waterproofing, windows. We have yet to find a respondent who could not run the self-prompting questionnaire.
Administering the program
There are several ways to administer the conjoint analysis program. Personal interviews are the logical first choice, but are expensive. We have had success with mailed diskettes and having trade show attendees fill out the programmed questionnaire. Each is much lower in cost and yields comparable results.
Mailed diskettes: The ACA diskettes are ideal for mailing. Prospective respondents first are contacted by telephone to be certain that they are a critical part of the specifying/purchase decision process and that they are willing to participate in the study. In our experience, among respondents who are part of the decision process, 90 percent agree to cooperate. Access to a PC rarely is a problem; refusals most often are based upon lack of time.
The diskette is mailed and the respondent goes through the ACA program, returning the diskette in the mailer provided. Of the 90 percent who agree to participate, we find that two-thirds or more will complete the interview and return the diskette for an overall response rate of 60 to 70 percent.
Trade shows: Respondent administered, computerized questionnaires can be very effective at trade shows. In one instance 150 well-qualified attendees at a construction products trade show were interviewed using four PCs set up directly on the trade show floor. All of the interviews were conducted over two six-hour periods.
Results
Conjoint analysis studies yield valuable trade-off information on subjects where respondents might show unintended positive or negative bias when asked a direct question. Price/value relationships are a prime example. Business-to-business respondents tend to understate the importance of price. The various ways the price subject is approached in conjoint trade-off questions wipes out any bias, resulting in a true reading.
Another example is comparing the importance and value of technical product features. Engineers tend to fall in love with innovative ideas, a potential trap for market researchers. Conjoint analysis places the importance and value of specific features in perspective. An example we have seen several times is trying to reach a balance between precision performance of a product and durability/longevity.
Conjoint analysis seldom should be a stand-alone technique. Results are valuable, but most clients also need to understand the reasons behind the information obtained. The pilot study is helpful, but one-on-one personal and/or telephone interviews still are needed to develop a full understanding of the subject at hand.