While performing feature selection and pricing optimisation in one experiment may seem like a good idea, it may result in increased complexity and costs compared to multi-stage research.
Conjoint analysis is a powerful tool for performing feature selection and pricing optimisation. It is common to see researchers combine both their feature selection and pricing optimisation research into one large scale conjoint study. However, a multi-stage experiment to find the most preferred features first, and then optimising prices for the best features may reduce overall costs and improve the reliability of results.
To explain the benefits of multi-stage testing, lets look at the example of Brand Co, a manufacturer of liquid meals based in the US.
Case study: Optimising Brand Co’s Liquid Meals
Brand Co. was interested in launching a new line of healthy plant-based liquid meals in the US, and wanted to determine the best features and pricing for their new product. To do so, Brand Co. contacted Conjointly to run a Brand-Specific Conjoint with the following features:
- Flavour: 8 levels
- Format: 3 levels
- Size: 4 levels
- Logo: 3 levels
- Claims: 6 levels
- Price: every 50c from $6 to $12.
The experiment also included several prominent competitor products. The single stage study would require a minimum of 500 respondents, costing Brand Co. $2500 for the respondents. Brand Co. then asked Conjointly’s market research experts to review the study before launch.
The benefits of multi-stage testing
In the review, our team explained that a multi-stage approach would improve reliability and reduce costs. In a single stage study, the full set of features and pricing need to be explored. This can be visualised by the light green rectangle in the diagram below.
In a multi-stage experiment, the first stage tests the only the preferences of features using a Generic Conjoint. This significantly reduced the number of unique combinations covered in stage one, shown as the orange area in the diagram below.
Using the best features from stage one, the second stage can then perform pricing optimisation in a Brand-Specific Conjoint with a reduced set of features. This reduced feature set is visualised as the blue area in the diagram below.
Running feature selection and pricing optimisation in two stages reduces the total area of space explored in the study, leading to lower overall costs from fewer required respondents as well as more reliable results as each unique combination of features and pricing can be exposed to more respondents. Intuitively, by eliminating features that perform poorly respondents are not “wasted” on testing pricing for features that are not preferred in the first place.
How multi-stage testing helped Brand Co. reduce costs and improve reliability
After the consultation with Conjointly, Brand Co. decided to conduct a multi-stage study.
In stage one of their study, Brand Co. conducted a Generic Conjoint to perform feature selection requiring 200 respondents. Compared to the over 20,000 unique combinations in the single stage study, stage one only had around 1,700 combinations. This meant that even with the reduced number of respondents, each combination was exposed to more respondents.
After eliminating features that were not preferred, Brand Co. proceeded to the pricing optimisation stage with a reduced feature set
In stage two, Brand Co. performed a Brand-Specific Conjoint, including competitor products to optimise prices in a competitive context. Stage 2 required 250 respondents to cover 750 unique combinations. From stage two, Brand Co. was able to find the optimal pricing for their product.
By utilising multi-stage testing, Brand Co. reduced the total respondents required in their study from 500 to 450, a saving of $500. In addition, the improved reliability from multi-stage testing meant that Brand Co. was able to launch their product with more confidence.
