Conjoint analysis is used to assess how much value people place on specific features when making a purchase decision. The primary objective is to determine what combination of attributes associated with these various features is most successful in driving people to make a purchase decision.
For example, a health food company developing a protein shake may want to know what combination of features would be most desirable for a specific segment of their market. Features that make up the protein shake such as the number of grams of protein, the number of grams of carbs, the flavor, whether it’s organic or non-organic, and the price are measured against each other. This is done by presenting various attribute combinations that are associated with each feature to the consumer and asking them to pick which one they prefer as below.
Which of the following protein shakes would you be most likely to choose to purchase after a workout?
After selecting the best option, another similar scenario is presented with a slightly different mix, and then another one after that and so on.
Any number of features (protein, flavor, etc.) and number of attributes associated with each feature (chocolate, banana, strawberry, etc.) can be chosen. However, in practicality the number of features should not exceed five and the number of attributes (also know as levels) should not exceed four. The more attributes and features there are, the more scenarios the survey participant will have to fill out, so keeping them to a minimum is important so as to prevent participants from dropping out before they have completed the survey, or resorting to simplification techniques to simply get through it.
One of the main advantages of conjoint analysis is that it presents a scenario that more closely mirrors the actual choices consumers are presented in the marketplace. Asking customers to rate or rank different options may be useful, and preferred in some situations, but is not as realistic.
Conjoint analysis also allows for measuring how much difference each feature ads to the total utility of the product.
Conjoint analysis indicates how sensitive preferences are to changes in product features, and can uncover drivers that may not be apparent, even to the respondent.
One of the difficulties encountered with conjoint analysis is the tendency of survey participants to fixate on price/cost. If that appears to be the case, the study can be run without the price/cost. This will eliminate the ability to determine price sensitivity, but also takes care of the “price-fixation” factor. Another method can be to define the price/cost as a range – Eg. ($5.99 – $7.99). This can help turn the focus away from price/cost to the other attributes.
Additionally, because the number of attributes is limited, conjoint analysis does not give a very accurate picture regarding market share.
The applications of conjoint analysis are many. It can help determine what configuration of features will be most successful in the design of new products. It can help find out what affect a pricing change may have on sales. Conjoint analysis can be used to analyze trade-offs people are willing to make with regard to various service options.
Associated Statistics: There are several statistical measures associated with conjoint analysis some of the most important include:
Part-Worth Functions: The part-wroth describe the utility consumers attach to the attributes associated with each feature.
Relative Importance Weights: Indicates which attributes are important in influencing consumer choice.
Fractional Factorial Designs: Designs employed to reduce the number of stimulus profiles to be evaluated in full profile conjoint analysis
Orthogonal Arrays: Special classes of fractional designs that enable the efficient estimation of all main effects.
Choice-Based Conjoint (CBC)
Choice-Based Conjoint is the most popular type of conjoint analysis. CBC asks the respondents to choose from a set of concepts. It is primarily used to study the relationship between price and demand, and works best when only a few features need to be considered.
Adaptive Conjoint Analysis (ACA)
As mentioned earlier, one of the challenges facing conjoint analysis is that while technically there is no limit the number of attributes and levels, in practicality there is. Too many attributes creates a survey that is too long and cumbersome. The solution is Adaptive Conjoint Analysis (also called Self-Explicated). With ACA the survey is adapted for each respondent. Early in the process the computer learns what attributes are most important to the respondent and focus on those. Areas of little or no interest are eliminated. This can also lead to higher quality data as the respondents can be more interested in the questions.
Full-Profile Conjoint Analysis
This type of conjoint analysis uses a rating or ranking scale for each scenario. There are two approaches, Pairwise and Single-concept.
Which trail-running shoe would you prefer?
How likely would you be to purchase this trail-running shoe?
Pair-wise is best used when a comparing competing products. When acceptability is the primary concern, the Single-concept approach would be the preferred method.