Personal information about applicants is often requested by service providers to be used as an input of a classification process establishing the specific situation of each applicant. This is a prerequisite for the service provider to make an appropriate offer to the applicant. For example, the rate and duration of personal loans are usually adapted depending on the risk based on the income, the assets or past lines of credits of the borrower. In the eyes of privacy laws and directives, the set of exposed documents collected to achieve a service must be restricted to the minimum necessary. This Limited Data Collection principle reduces the impact of data breaches both in the interest of service providers and customers. In this article, we show that in practice, the data collected traditionally is excessive. We propose a new approach that we call Minimum Exposure, where the minimum set of documents required can be computed on the user's side. We formalize the underlying problem and show it is NP hard. We propose algorithms to compute a solution and validate them with experiments. The Minimum Exposure approach leads to a very significant reduction of the quantity of personal information exposed, therefore leading to important privacy gains for the applicant and large scale savings for service providers in the event of data breaches.