Geographic routing is becoming an attractive routing solution for WSNs since it offers a radical departure from traditional topology-dependent routing paradigms through use of geographic location in data delivery. However, it often suffers from the routing hole, referring to an area free of nodes in the direction closer to destination, in various real-world environments such as buildings and obstacles, leading to route failure. Currently, most geographic routing protocols tend to exploit face routing to recover the route. The basic idea behind it is to planarize the whole network by eliminating the crossing links before applying routing algorithms, thus achieving suboptimal network performance. In this article, we first survey representative face-based geographic routing approaches, including their design prerequisites and philosophy. Furthermore, we outline the emerging issues to be addressed in the future and illustrate the forming factors behind them in detail. Based on these observations, we then propose a CAGR to address the routing hole problem by employing relative coordinate systems, avoiding planarizing networks and preserving route optimality properties. Simulation results show that the proposed approach is superior to existing protocols in terms of packet delivery ratio, control overhead, and delivery delay in WSNs over a variety of communication sessions passing through the routing holes.