The last decade has seen the explosion of the Internet of Things (IoT), which is enabling a range of new applications based on the connection of physical objects to the Internet. The growing diversity of IoT connectivity technologies is bringing new challenges to IoT solution designers. Indeed, it is increasingly difficult to choose and configure the network technology for a given use-case. In this article, we formalize and investigate the design optimization problem for selecting and configuring the IoT connectivity technology of an application that can evolve over time. Finding the right abstractions and the good balance between performance and evaluation complexity to compare networking options is a key research challenge. To address this problem we propose to separate the concerns of IoT application architects from those of network experts and to provide a methodology, HINTS, to help designers in making customized decision. HINTS combines IoT application requirements and goals abstraction, IoT network modeling, discrete-event network simulation and a multiple attribute decision making method. The application of the methodology on three use-cases highlights how it helps in (i) selecting the best network technology option, (ii) defining an appropriate configuration and (iii) anticipating the behavior when device density or workload intensity scales up. The main contribution of this paper is to propose the first formal approach and associated algorithm to automatically optimize the design of the IoT connectivity of an application. Results show that it can yield up to a factor two improvement in the solution performance.