Menu-Based Choice (MBC) is a variant of conjoint analysis for products where customers assemble a basket of components rather than pick one alternative from a set. Standard CBC asks “which of these three cars would you buy?” — MBC asks “which of these 12 add-ons would you include with your car, and at what price?”
How it works
Each menu item is modelled as its own binary take-or-leave decision. The probability of respondent $n$ taking item $i$ is a logistic function of its utility relative to a per-respondent “take-it” threshold, with an optional price term:
$$P(\text{take } i \mid n) = \frac{1}{1 + \exp(-(u_{ni} - \theta_n + \gamma_n p_i))}$$
where $u_{ni}$ is respondent $n$’s utility for item $i$, $\theta_n$ is their threshold (how easily they say yes), $\gamma_n$ is their price sensitivity and $p_i$ is the item’s price. Utilities, thresholds and price coefficients are all estimated hierarchically.
When to use MBC
Modular products — software with configurable modules, insurance plans, loan options with add-ons
Fast-food menus — which add-ons, upsells and sides customers combine
Streaming / subscription bundles — which channels or packs customers mix
Automotive configurators — which options customers add to a base vehicle
Service tier design — which features belong in which pricing tier
MBC vs CBC
Use CBC when customers choose between discrete alternatives (buy brand A OR brand B). Use MBC when customers combine multiple components (buy A and B and C, but not D). The two generate different data structures and require different estimators.
What MBC tells you
Take rates for each item at each price
Cross-item effects — does taking A make B more or less likely?
Portfolio optimisation — which combinations maximise basket value
Price elasticity per item — how take rate responds to price changes
Bundle pricing — what to charge for a pre-packaged combination