Practicing chemists typically try to predict whether or not an instance of a particular generalized reaction will yield product based on the molecular connectivity alone. Such predictions based on connectivity are generally considered plausible, but the physical outcome of the actual reaction is ultimately dependent on a number of factors that are practically impossible to predict to any useful accuracy (solvent, atomic orbitals etc.).
As such, chemoselectivity can be difficult to predict but observing selective outcomes in cases where many reactions are plausible, is common. Examples include the selective organic reduction of 4-nitro-2-chlorobenzonitrile to the corresponding aniline, 4-amino-2-chlorobenzonitrile and the greater relative chemoselectivity of sodium borohydride reduction vs. lithium aluminium hydride reduction.