Pricing something that had never existed before
How I used conjoint, willingness-to-pay analysis, and market simulation when there was no historical price data to work from.
Method: Choice-based conjoint, Hierarchical Bayes, logit modeling, willingness to pay, interaction effects, market simulation, bundle analysis, and price optimization.
What we needed to figure out
The team was preparing a new consumer product, which meant we had no sales history to fall back on. We still needed to decide which features people cared about, what they would pay, how the offering should be packaged, and whether the different products would compete with one another.
What I did
I owned the study from the first research plan through the executive readout. After weighing the options, I chose choice-based conjoint. Our survey platform did not support conjoint directly, so I had to get creative about how to field it. I built the choice tasks outside the platform, turned them into images, and collected the answers through standard survey questions.
How I approached the analysis
Once the data was cleaned and reorganized, I used a Hierarchical Bayes model for the conjoint estimation. I then worked through attribute importance, utility, willingness to pay, price sensitivity, and feature-price interactions. The product team also wanted to understand market share, bundling, cannibalization, and the revenue-maximizing price, so I extended the work into market simulation and price optimization.
The analysis was technical, but the audience did not need a statistics lecture. A large part of my job was explaining why the method fit the problem, what the assumptions were, and how the team should use the result. I iterated on that explanation until the reasoning was clear enough to support an actual decision.
What the work changed
The product team and company leadership used the research to choose the offering, packaging, and price. It replaced a decision that would otherwise have relied heavily on guesswork with a clear model and a set of tradeoffs everyone could see.