Adsplit: A multi-brand advertising budget allocation model

ADSPLIT is a computer-based, interactive, marketing model which optimally allocates a specified corporate promotional budget among individual brands competing for limited resources. The model requires either (a) regression-based response function coefficients based on historical data, or (b) judgments on what sales should be for different values of price and advertising expenditure for the brands, which the program then uses to estimate the function parameters. The advertising-sales response function is modeled through a flexible form that allows both concave and S-shaped relationships, while sales are related to price through a constant elasticity (Cobb-Douglas) function. The model uses the estimated or input functions to compute the optimal budget allocation for the brands, given an upper bound for the total promotional expenditures and upper- and lower-budget bounds for every brand, using nonlinear optimization heuristics. Optimal prices for the brands are also computed.