Merchants often use marketing elements such as advertisements, coupons and product recommendations, to attract customers and to convert visitors to buyers. We present a model for making a series of recommendations during a customer session. The model comprises of the customer’s probability of accepting a marketing element from a marketing spot and a reward for the marketing element. The probabilities can be estimated from customer history (such as traversals and purchases), while the reward values could be merchant specified. We propose several recommendation strategies for maximising the merchant’s reward and analyse their effectiveness. Our experiments indicate that strategies that are dynamic and consider multiple marketing spots simultaneously perform well.