Course · 2. Choice-based conjoint (CBC)

The optimizer

The simulator answers "what if"; the optimizer answers "what is best". You fix the competitors, say which of your product's levels may change and over which prices, and choose a goal.

Goals

• Most share: the version that wins the most choices.
• Most revenue: share × price, per shopper making a choice.
• Most profit: share × (price − unit cost). Enter a unit cost and, if some levels cost more to make (a bigger can, a premium flavor), the extra cost per level. Multiply by your number of shoppers for totals.

Profit is where getting switching right pays off: the best price depends on how many buyers you lose to each rival as you raise it, and on how much of a new product's volume is taken from your own range.

How it searches

Chosŏn simulates 20,000 shoppers from the fitted model, each with their own tastes and in-the-moment preferences (products that share a flavor share that part, as the switching structure says), and scores every candidate on all of them. That is fast enough to try thousands of combinations. The best ten are then scored again with the exact model, with 95% ranges, and Chosŏn reports the chance that the first really beats the second. When that chance is low, the top options are close and any of the first few is defensible: pick on what the survey does not cover (production, brand fit).

With several of your own products, the optimizer improves one at a time, holding the others, until nothing changes.

Limits

Prices stay within the range tested in the survey. Costs are per unit and do not include fixed costs. Shares are shares of choices on the shelf you describe, not market shares.

Next: When our model helps