In order to facilitate the execution of repetitive tasks, program synthesis approaches have been developed. Program synthesis consists in automatically inferring programs that satisfy a user's intention. FlashFill, in Excel, is the most famous. In FlashFill the user's intention is represented by examples, i.e. pairs (input, output). FlashFill explores a very large search space of programs and thus may require a large runtime and infer many programs, some of which work on the given examples but do not capture the user's intent. In this paper we present Quickfill, an approach that reduces the space of programs to infer by enriching the user specifications. We show that with this approach, it is often possible to give fewer examples than with the original FlashFill algorithm, and that the proportion of correct programs obtained is greater in most cases.