The global Internet of Everything entails the collection and dissemination of information throughout the entire Earth, demanding the massive deployment of Internet of Things (IoT) devices. Recently, LEO constellations of nano and small satellites have been presented as a strong candidate for providing global connectivity, with the Direct-to-Satellite IoT (DtS-IoT) architecture as a promising solution. However, DtSIoT connections may increase exponentially, representing a great scalability challenge to overcome in future deployments. We defend the hypothesis that prior knowledge of the number of competitors contending for a shared channel contributes to the efficient allocation of temporal and spectral resources. For this purpose, in this work, we present and adapt a collision set size estimator for DtS-IoT networks, which is intended to provide feedback to a MAC protocol operating over the satellite link. Our initial results show that the accuracy of the estimator depends on the number of nodes deployed and the size of the transmission frame, as well as the distribution of the IoT nodes on the ground. As expected, sensors uniformly distributed in the area involve smaller estimation errors than deployments following geographical clusters. Likewise, when the number of slots in the frame approaches the number of contending nodes, the estimation error decreases. Therefore, a trade-off between the estimator accuracy and the number of resources allocated to communication is established, which must be compensated with the efficient use of resources by MAC protocols based on the estimation information.