Conjoint analysis is one of the most popular methods used to measure the preferences of individuals or groups. It determines, for instance, the degree to which consumers value specific products, which then helps in developing sound pricing strategies. However, this method has some pitfalls that can hinder its effectiveness and application in pricing research, which the author addresses in this article. Dr. Felix Friederich is a Business Development Consultant at Horváth Germany. He can be reached at FFriederich@horvath-partners.com.

The Pricing Advisor, November 2023

Introduction

In the realm of pricing strategies, the application of robust research methodologies stands as a cornerstone for informed decision-making. Discrete choice experiments, also called conjoint analysis, have been widely applied in various industries to understand consumer preferences and inform strategic decision-making, particularly in pricing. Whether it’s Toyota and Ford in the automotive sector, Apple and Samsung in the technology industry, or Procter & Gambler and Unilever in retail and consumer goods, they all use this method to make better pricing decisions.

However, while the allure of conjoint analysis in pricing remains undeniable, a critical caveat demands attention. The effectiveness of the method depends on careful adherence to certain criteria and stringent standards that many practitioners still stumble over. In fact, failing to fulfill these requisites not only undermines the effectiveness of the method but also risks distorting the insights derived, potentially leading to flawed pricing decisions that fail to resonate with the targeted consumer base. This article identifies five pitfalls of conjoint analysis to help practitioners successfully and effectively use this method to optimize their pricing strategies.

Conjoint Analysis in Pricing Research

Conjoint analysis is a quantitative research method used to measure how consumers value different product attributes or features.^[1] It estimates the utility or value that consumers associate with each attribute. Specifically, conjoint analysis is highly effective in pricing research due to its ability to reveal consumer preferences and willingness to pay for specific product attributes. By assessing how consumers prioritize and trade off different attributes against price, the method helps in understanding the price sensitivity of consumers and in determining the optimal pricing strategy that maximizes perceived value and market competitiveness. The major advantage of the method lies in the simulation of a real purchase scenario, as participants have to choose between different options instead of ranking them. This makes the method a valuable tool for developing pricing strategies and optimizing product offerings to meet consumer demand. However, reaping the benefits of this method depends on several pitfalls that practitioners often stumble upon. The following section therefore outlines the five most common pitfalls of conjoint analysis.

The Five Pitfalls of Conjoint Analysis

Research scope definition

The first pitfall relates to the scope, or the objectives, it is used for. Conjoint analysis is particularly useful in determining the optimal combination of product features, functionalities, and consumer price sensitivity. Yet, for certain research objectives, other traditional methods might be more applicable due to lower complexity and scope focus. For instance, for pricing research aiming to solely identify willingness to pay or price elasticities, the Price Sensitivity Meter (PSM) or Gabor Granger are useful. In specific, both methods are more suitable for quick market assessments as they are more practical and less complex in the study design. Second, both methods are specifically useful when the primary focus lies in assessing consumer price sensitivity as a single attribute, without evaluating trade-offs or preferences for multiple attributes simultaneously. Moreover, in resource-constrained situations (e.g., data availability) these methods are more likely to yield effective results without the need for extensive data collection and analysis. On the other hand, neuroscience tools like EEG, fMRI, and eye-tracking provide a deeper understanding of consumer preferences. Specifically, neural responses shared across individuals represent a universal index of preference and can be used to improve forecasts of market behaviour, particularly in terms of price perceptions.

Therefore: Pricing researchers should not lose focus of the research scope. Conjoint analysis is helpful in exploring consumer utility perceptions, but it is not an “all-purpose” research method!

Capabilities and resources

The second pitfall relates to the research capabilities. Projects can fail based on unfavorable decisions of executing in-house research or engaging external vendors. When considering in-house pricing research, the resources in terms of expertise and specialized skills to effectively conduct such projects have to be carefully considered. For instance, you should evaluate if access to necessary tools and technologies is available. That is, pricing researchers need to ensure the availability of typical purpose or statistical programming languages (e.g., Python, R or SPSS) as well as the handling of multinomial logistic or simple logistic regressions. Furthermore, practitioners need access to survey or feedback management tools (e.g., Qualtrics or Open Lab) for data collection. Conversely, when hiring external vendors, evaluating the vendor’s track record and past performance in conducting pricing research projects is critical. Furthermore, assessing the efficiency and speed of execution that vendors can offer, taking into account their expertise, established methodologies, and dedicated resources for conducting pricing research is essential. Finally, the provision of in-depth industry knowledge and market experience from external vendors is crucial. This applies in particular to understanding specific industry-relevant product features and pricing schemes.

Therefore: Research projects should not fail due to a lack of internal capabilities or collaboration with vendors who lack industry expertise – rigorous evaluation is key!

Attribute and level selection

The third pitfall relates to the attribute and level selection. An unfavorable choice of attributes and corresponding levels leads to mediocre study insight. Therefore, well-chosen attributes and levels help in capturing the most relevant information from respondents, accurately reflecting consumer preferences. First, the selected attributes and levels should be realistic and practical, aligning with the actual product characteristics that consumers encounter in the market. Unrealistic or overly complex attributes may lead to respondent confusion and compromise the validity of the study. Second, achieving balance and orthogonality in the design of attributes and levels is essential. That is, to minimize potential bias and ensure that each attribute contributes independently to the overall utility function. And third, conducting pilot tests to validate the relevance of the selected attributes and levels is critical before proceeding with conjoint analysis. With this, the overall suitability of the attribute set among the target respondent group can be assessed. For instance, pre-tests with small sample sizes ranking the attributes and levels importance are an uncomplex and fast validation approach.

Therefore: The importance of an appropriate set of attributes and corresponding levels is often neglected but is key for insightful results. Resources must be devoted to this topic!

Design generation

Once the proper attributes and levels have been selected, the fourth pitfall lies in the optimal design of the conjoint analysis to produce reliable and valid results. The design of such analysis will determine to a large extent the accuracy with which the preference parameters can be estimated. First, the design should carefully distribute attribute levels to ensure that each attribute contributes independently to the overall consumer preference estimate. This helps in preventing bias and confounding effects that may arise from uneven distributions or interactions among the attribute levels. Second, pricing researchers want to avoid respondent fatigue. That is, a well-structured design with a balanced number of choice sets and a manageable number of attributes helps in sustaining respondent interest and minimizing the likelihood of dropouts or incomplete responses. Third, a robust design ensures efficient estimation of preference utilities and partworths. This leads to more accurate and reliable results. By optimizing the design structure and maximizing the information content of the choice sets, researchers can enhance the statistical efficiency of the analysis and improve the precision of the preference estimates. When engaging with external vendors, the design creation and certain threshold applied for optimal design set-up should be evaluated.

Therefore: An inefficient design leads to unreliable results. This pitfall should be avoided by carefully generating optimal study designs!

Sample selection

The final pitfall relates to the sample size. Sampling plays an essential role in producing insightful and useful results, but it is often a victim of cost-cutting in studies. To avoid obtaining mediocre results, several aspects have to be considered. First, a well-defined and representative sample ensures that the results accurately reflect the preferences and behaviours of the target population. It helps in generalizing the research results to the broader market, enabling more reliable and applicable insights for decision-making. For this, the sample should not be restricted (ex-ante) to certain demographics or psychographics (rather using sample diversity ex-post). Second, the size and composition of the sample directly impact the statistical power of the study, and thus, the results. A sufficient sample size (~300 to 400 participants) enhances the statistical power of the analysis, allowing the detection of relevant differences in preferences and accurate estimates of preference parameters. This, in turn, ensures the robustness and reliability of the research findings.^[2] Finally, a robust sample size allows for segmentation analysis (also in interplay with sample diversity) which allows identifying distinct consumer segments based on their preferences for specific product attributes and prices.

Therefore: The sample size should not fall victim to the “reduction of study costs,” as this directly leads to poor results. To avoid this pitfall, a robust sample size is required!

Conclusion

Conjoint analysis is a powerful and insightful tool for conducting pricing research. By assessing consumer preferences and trade-offs across various product attributes, it offers valuable insights into willingness to pay and price sensitivity. However, in order to fully exploit the potential of the method and derive meaningful pricing decisions, the pitfalls outlined in this article must be taken into account. In this way, the method’s potential can be fully leveraged to help practitioners make informed pricing decisions and ultimately increase revenues.

1. The difference between conjoint analysis and discrete choice experiment is not considered. Traditional conjoint analysis is based on conjoint measurement, while discrete choice experiments are based on random utility theory. ↑
2. Sample size depends to a large extend on number of included attributes and levels. ↑