Nowadays, e-commerce has enabled tourists to plan and purchase holidays more efficiently. Therefore, an efficient service provider needs to gain insight into the behaviour of potential tourists in the pre-purchase phase so as to better target promotions and move customers to purchase. Using clickstream data on the traceable and anonymous users of a tourist accommodation services website, our paper explores their online journey by considering browsing profiles as predictors of purchasing probability. We perform a two-stage analysis: first, identifying browsing profiles by means of a mixture hidden Markov model, and second, investigating purchase probability using a logit model based on users’ website history. We apply this approach to data from Sunweb, a Dutch online holiday provider.