Although the transfer of professors’ inventions is typically performed by an intermediary set up by the university (the technology transfer office), other forms of transfer do coexist in reality. To clarify this situation and its consequences, we develop a model that endogenizes a professor’s decision regarding the form of transfer for her invention in which intermediation by the transfer office is only one of two options, the other one being a transfer carried out by the professor herself. The intermediary can reach more potential licensees of the invention, whereas the professor is usually better at mitigating information asymmetry through signaling. In the semiseparating equilibrium of the game, promising technologies are likely to be cherry‐picked and transferred by the professor. This prediction that the transfer mode is influenced by invention quality is tested and confirmed, as are several other predictions, on our sample of 446 UK academic inventors. Specific care is taken to control for other mechanisms of IP assignment, such as those derived from consulting and sponsored research. Insights in terms of policy are drawn from this new perspective focusing on alternative transfer channels, match quality, and information asymmetry rather than on moral hazard.