Personalized medicine (PM) is notably typified by the development of companion diagnostic tests to guide optimal treatment selection. PM has thus the potential to dramatically improve patients’ outcomes and optimise allocation of resources. However, very few attempts exist that transparently include diagnostic test performance such as Sensitivity (Se) and Specificity (Sp) into cost-effectiveness and budget impact models. This research proposes an analytical framework to unveil diagnostic added-value according to different diagnostic performance scenario. The framework is based on a decision tree and compares two hypothetical treatments N vs. C. N is a new treatment associated with a companion diagnostic test T. C is the current standard of care not associated with any test. T selects the likely responding patients based on the presence (T+) or absence (T-) of a predictive sign of response to N (a distinctive biomarker for instance). We demonstrate that it is the prior prevalence of the sign within the target population coupled with the expected effectiveness differential between N and C in true positive patients and with the performance of the test (Se and Sp) that are the fundamental determinants of the potential value of a PM strategy. An extension of the model to the case of 2 competing PM strategies (and thus 2 competing tests) is shown. We conclude that companion diagnostic test performance is key to achieve the promises of PM. This analytical framework allows payers, HTA bodies and manufacturers to gauge the potential value and financial impact of a PM strategy at all stage of its development.