Basically, genotyping is tumor analysis coupled with clinical trial literature search, which tries to match therapies to patient-specific biomarker information to generate a treatment approach. In other words, information that may help when considering "potential" treatment options (theoretical analysis).
Genotyping will only be of value for drugs for which a gene mutation is informative -- basically KRAS and EGFR mutations -- and we already have tests for those and they are of very limited value. Otherwise, it's a ton of information for which the drug selection value is pretty darn useless.
Phenotyping is tumor analysis that "actually" measures the response of the tumor cells to drug exposure. Following this exposure, it measures both cell metabolism and cell morphology. The integrated effect of the drug on the whole cell, resulting in a cellular response to it, measuring the interaction of the entire genome (actual analysis).
What's more important than what genes are in the DNA is what genes are actively making RNA, which RNA is actively making protein, which protein is being turned off or turned on, and how all of the proteins in the cell are interacting with each other.
The only way to get the latter information, which is ultimately what you want, is to treat the patient with phenotype analysis. In drug selection, phenotype analysis doesn't dismiss DNA testing, it uses all the information, measuring the interaction of the entire genome, to design the best treatment for each individual.
I believe this and other recent announcements have been overreached and giving us overexpectations. The research is exciting, but the path remains long and arduous. It is important to recall that despite the fact this is incredibly sophisticated and difficult work, it is still reasonably early in our ability to perform the analyses, interpret the data, and determine the best way to apply it to the clinic.