Effective distribution using collaborative fulfillment networks requires coordination among the multiple participating firms at different stages of the supply chain. Acting independently, supply chain partners fail to weigh the cost burden they impose on upstream suppliers when their replenishment order quantities vary from period to period. This paper explores a new approach to coordinate multiple stages in the supply chain by controlling, through appropriate downstream inventory management, the demand variability that is propagated to upstream stages. We propose and analyze a coordinated inventory replenishment policy that uses "order smoothing" to reduce order-size variability and thus reduce overall system costs, including both inventory and transportation costs. We characterize the optimal parameter values for smoothing alternatives (such as exponential smoothing and moving weighted average policies), assess their economic benefits, and develop insights regarding supply chain contexts that might benefit most significantly from reducing the variability of orders to upstream stages. Using the distribution network for specialty brand appliances as an illustrative example, we demonstrate the potential cost savings that order-smoothing strategies can yield compared to the uncoordinated case when individual firms separately minimize their costs. The magnitude of savings depends on several factors, including the variability in consumer demand, level of product variety, and degree of inventory aggregation in the distribution system. Based on our analytical results, we develop a framework to assess cost reduction opportunities through variability control for different supply chain scenarios.