Conjoint Analysis has been used for decades to evaluate and validate new products and improve existing products. Conjoint analysis helps you to choose the best feature and price trade-offs to gain market share against competitors.
Survey participants are shown alternative versions (generated by computer) of a product, with different features, prices and competing brands (if applicable). Survey participants are asked to evaluate each alternative. Enough alternatives are shown to estimate the impact of different features and pricing on preference.
These estimates of impact take the form of a statistical model, with each attribute level (feature), price and brand having a parameter (also called a part worth) assigned. The parameters indicate the impact of alternative attribute levels, pricing and brand and are useful for understanding what is important. It can also be used to segment the market by need. This can be taken a step further by creating a market simulator, allowing different scenarios to be tested.
How does conjoint analysis work?
Survey participants in the relevant market are asked to evaluate alternative product or service profiles. The alternatives are created based on a computer generated experimental design. The survey doesn’t need to test every possible alternative, since the goal is parameter estimation, rather than direct evaluation of the profiles shown. Once the survey is completed, the model parameters can be estimated. These parameters indicate the impact of brand, features and pricing on claimed purchase likelihood (alternative measures can be used if so desired). The mathematical model can then translate purchase likelihood into a preference share. Preference shares can be simulated for any particular brand, feature and price combination in order to test different scenarios.
How accurate is conjoint analysis?
Conjoint analysis can be surprisingly accurate. For instance a study in the cosmetics market by Acentric yielded a Mean Absolute Error (MAE) of 3.6% when estimating market shares in the current market; while a study of three FMCG categories by Orme & Heft (1999) yielded an MAE of 2.87%. That said there are examples of dismal performance as well; which is why it is important that careful attention is paid to study design and execution.
Who is conjoint analysis for?
Conjoint analysis is relevant to a wide range of organizations; and is useful in evaluating both products and services. Roles may include - but are not limited to - engineers, R&D / NPD specialists, developers, marketers, entrepreneurs and executives.
- Estimate potential share (preference share) for your product and competitor products (if you have no competitors, other options are available). Preference share is a function of brand equity, product features and pricing and is isolated from external factors (such as awareness and distribution levels) that may vary across competitors and time.
- Estimate preference share for products that don’t exist yet, or to improve existing products.
- Identify the best combinations of features, or compromises that improve results at acceptable cost.
- Determine the relative importance of different product attributes without asking consumers directly.
- Run scenarios in a simulator to see if your changes would have a positive or negative impact.
- Run 'what if' scenarios regarding competitor responses.
- Obtain a rough idea of the impact of price changes on share of preference, without needing to run in-market experiments that may harm your brand.
- Understand approximately what price change may compensate for feature loss, or what increases may be possible with feature additions to remain competitive.
- Obtain an indication of your brand's equity, by running a 'what if' simulation... what if all competing brands had the same features?
Why Acentric's CVA conjoint?
- Unlike traditional CVA conjoint analysis (e.g. IBM SPSS) that uses fixed decision rules (such as BTL) to translate rating or ranking scales into shares; a flexible linking function is used to emulate choice probabilities that adapts the sensitivity of choice probabilities to ratings, allowing for greater accuracy.
- It works better with small screens (e.g. smartphones).
- Simulators include demographic weights for representation.
- Simulators include individual purchase volume weights (necessary in situations where repeats of the same behaviour will occur in the time period of interest).
- More attributes can be accommodated with Acentric's Hybrid CVA Conjoint Analysis. Traditionally only 6 or 7 attributes could be accommodated.
- The model is estimated at the individual level, meaning part-worths can be used to segment using cluster-analysis. A-priori demographic segments can also be used in the simulator.
Read the Acentric blog article "Seven reasons to choose CVA over CBC conjoint analysis" for more insights.
Approximately 20 to 30 working days depending on the target group, questionnaire length and analysis complexity. Note: timelines are provided as a guideline only. Timelines are calculated from the date you approve the final questionnaire for launch.
View the conjoint simulator explainer video below for more insight into the optional simulator deliverable.