We develop a multi-period, multi-product Mixed Integer Linear Programming (MILP) model for omni-channel retailer that operates distribution centers (DCs) and stores and serves both walk-in and online demand. In each period, the firm decides how much to order into DCs, how much to replenish to stores, and how to fulfill online orders—shipping from DCs, shipping from stores, or BOPIS (buy-online-pick-up-in-store). The objective is to minimize total cost, including fixed ordering, unit purchasing, inventory holding, DC-to-store transport, online order delivery or BOPIS handling, and reverse-logistics costs from product returns. The model unifies lot-sizing, split fulfillment (means an online demand is served by more than one location, for example, some units ship from a DC, some ship from store, or BOPIS), and returns, informing tactical planning and policy evaluation for omni-channel retailing practice. A test dataset validates the formulation: it lowers total cost, balances store workload, and recovers inventory versus a DC-only benchmark.