The Colorectal Cancer Subtyping Consortium identified four gene expression consensus molecular subtypes, CMS1 (immune), CMS2 (canonical), CMS3 (metabolic), and CMS4 (mesenchymal), using multiple microarray or RNA-sequencing datasets of primary tumor samples mainly from early stage colon cancer patients. Consequently, rectal tumors and stage IV tumors (possibly reflective of more aggressive disease) were underrepresented, and no chemo- and/or radiotherapy pretreated samples or metastatic lesions were included. In view of their possible effect on gene expression and consequently subtype classification, sample source and treatments received by the patients before collection must be carefully considered when applying the classifier to new datasets. Recently, several correlative analyses of clinical trials demonstrated the applicability of this classification to the metastatic setting, confirmed the prognostic value of CMS subtypes after relapse and hinted at differential sensitivity to treatments. Here, we discuss why contexts and equivocal factors need to be taken into account when analyzing clinical trial data, including potential selection biases, type of platform, and type of algorithm used for subtype prediction. This perspective article facilitates both our clinical and research understanding of the application of this classifier to expedite subtype-based clinical trials.