We consider an online assortment optimization problem with n substitutable products with reusable capacities. In each period, a user with a preference model arrives to the platform, and is offered an assortment of products. The user selects a product p from the offered assortment according to her choice model and uses it for a random number of periods Tp, where Tp is i.i.d. distributed based on some distribution depending only on p. Each usage of product p generates a revenue (Rp) for the platform. The goal is to compute a policy that maximizes the expected revenue of the platform over a finite horizon. In this talk we show that a simple myopic policy, which does not require any information about future arrivals or the distributions of usage time, provides a good approximation with respect to a clairvoyant optimal that has full information about the sequence of user types (or choice models) and the usage time distributions. In particular, we show that our myopic policy is 1/2-competitive. Our analysis is based on a coupling argument that allows us to bound the expected revenue of the optimal algorithm in terms of the expected revenue of the myopic policy.