G protein-coupled receptors (GPCRs) are essential cell membrane signaling molecules and represent the most important class of drug targets. Some signaling pathways downstream of a GPCR may be responsible for drug adverse effects, while others mediate therapeutic efficacy. Biased ligands preferentially activate only a subset of all GPCR signaling pathways. They hold great potential to become next-generation GPCR drugs with less side effects due to their potential to exclusively activate desired signaling pathways. However, the molecular basis of biased agonism is poorly understood. GPCR activation occurs through allosteric coupling, the propagation of conformational changes from the extracellular ligand-binding pocket to the intracellular G protein-binding interface. Comparison of GPCR structures in complex with G proteins or beta-arrestin reveals that intracellular transducer coupling results in closure of the ligand-binding pocket trapping the agonist inside its binding site. Allosteric coupling appears to be transducer-specific offering the possibility of harnessing this mechanism for the design of biased ligands. Here, we review the biochemical, pharmacological, structural, and biophysical evidence for allosteric coupling and delineate that biased agonism should be a consequence of preferential allosteric coupling from the ligand-binding pocket to one transducer-binding site. As transducer binding leads to large structural rearrangements in the extracellular ligand-binding pocket, we survey biased ligands with an extended binding mode that interact with extracellular receptor domains. We propose that biased ligands use ligand-specific triggers inside the binding pocket that are relayed through preferential allosteric coupling to a specific transducer, eventually leading to biased signaling.