A compound that binds its target tightly can still fail as a drug. Binding is only the first of several requirements, and they are largely independent of one another.
A candidate must reach its target in the body, which means surviving the journey from the gut into the bloodstream (absorption), avoiding rapid breakdown by liver enzymes (metabolism), and being cleared at a rate that keeps useful concentrations for hours rather than minutes (excretion). It must act on the intended protein without strongly engaging unrelated proteins (selectivity), because off-target activity is a common source of side effects. It must not be toxic on its own or through the breakdown products it forms. And it must be practical to manufacture at scale.
These requirements are not aligned. A molecule can be an excellent binder and still be destroyed by the liver within minutes, or bind its target and a heart-ion-channel protein with similar strength. Chemists summarize the situation as multi-parameter optimization: the goal is not the best value on any single axis but a set of values that is acceptable on all of them at once. A radar plot makes this visible — each axis is one property, and a compound is only viable if its shape stays outside the minimum acceptable ring on every axis.