Remarkable progress has been made in the development of biomarker-driven targeted therapies for patients with multiple cancer types, including melanoma, breast and lung tumours, although precision oncology for patients with colorectal cancer (CRC) continues to lag behind. Nonetheless, the availability of patient-derived CRC models coupled with in vitro and in vivo pharmacological and functional analyses over the past decade has finally led to advances in the field. Gene-specific alterations are not the only determinants that can successfully direct the use of targeted therapy. Indeed, successful inhibition of BRAF or KRAS in metastatic CRCs driven by activating mutations in these genes requires combinations of drugs that inhibit the mutant protein while at the same time restraining adaptive resistance via CRC-specific EGFR-mediated feedback loops. The emerging paradigm is, therefore, that the intrinsic biology of CRC cells must be considered alongside the molecular profiles of individual tumours in order to successfully personalize treatment. In this Review, we outline how preclinical studies based on patient-derived models have informed the design of practice-changing clinical trials. The integration of these experiences into a common framework will reshape the future design of biology-informed clinical trials in this field.