Bounds and Heuristics for Multi-Product Personalized Pricing

We present tight bounds and heuristics for personalized, multi-product pricing problems. Under mild conditions we show that offering a non-personalize price in the direction of positive vector (factor) has profit guarantee relative to optimal personalized pricing. An non-personalized is choice factor, when known. Using factor with equal components results uniform exceedingly sufficient bound hold. A robust presented achieves best possible performance guarantee. As an application, our model yields lower-bound on linear non-linear pricing, suggests effective solutions. Additionally, provides guarantees simple strategies such as bundle-size component-pricing respect mixed bundling policies. Heuristics cluster customer types are also developed goal improving by allowing each along its own factor. Numerical variety demand models, factors clustering heuristics. In experiments, economically motivated coupled machine learning performed best.