This work aims to sort cocoa beans according to chocolate sensory quality and phenolic composition. Prior to the study, cocoa samples were processed into chocolate in a standard manner, and then the chocolate was characterized by sensory analysis, allowing sorting of the samples into four sensory groups. Two objectives were set: first to use average mass spectra as quick cocoa-polyphenol-extract fingerprints and second to use those fingerprints and chemometrics to select the molecules that discriminate chocolate sensory groups. Sixteen cocoa polyphenol extracts were analyzed by liquid chromatography–low-resolution mass spectrometry. Averaging each mass spectrum provided polyphenolic fingerprints, which were combined into a matrix and processed with chemometrics to select the most meaningful molecules for discrimination of the chocolate sensory groups. Forty-four additional cocoa samples were used to validate the previous results. The fingerprinting method proved to be quick and efficient, and the chemometrics highlighted 29 m/z signals of known and unknown molecules, mainly flavan-3-ols, enabling sensory-group discrimination.