All hotels receive many booking requests every day, but only a small percentage of these requests are converted into reservations. Low conversion rates generate an additional layer of uncertainty into the hotel demand function and pose a challenge for revenue maximization. This study aims to investigate the conversion rate of hotel room sales in order to find the key variables that affect it. The hotel selected for the analysis is located in a well-known Italian summer resort that is open year-round. The data set consists of a collection of room requests made by customers through the hotel's website. To study the determinants of conversion rate, a logistic regression was performed using clusters of stay dates. Results provide information about the variables influencing the conversion rate for different periods of the year. These analytical insights generated with machine learning models might help practitioners to develop tools and strategies to maximize revenue.