The LC-MS based metabolomics studies are characterized by the capacity to produce a large and complex dataset being mandatory to use the appropriate tools to recover and to interpret as maximum information as possible. In this context, a combined partial least square discriminat analysis (PLS-DA) and two-way hierarchical clustering (two-way HCA) using Bonferroni correction as filter is proposed to improve analysis in human urinary metabolome modifications in a nutritional intervention context. After overnight fasting, 10 subjects consumed cocoa powder with milk. Urine samples were collected before the ingestion product and at 0-6, 6-12, 12-24 h after test-meal consumption and analysed by LC-Q-ToF. The PLS-DA analysis showed a clear pattern related to the differences between before consumption period and the other three periods revealing relevant mass features in this separation, however, a weaker association between mass features and the three periods after cocoa consumption was observed. On the other hand, two-way HCA showed a separation of four urine time periods and point out the mass features associated with the corresponding urine times. The correlation matrix revealed complex relations between the mass features that could be used for metabolite identifications and to infer the possible metabolite origin. The reported results prove that combining visualization strategies would be an excellent way to produce new bioinformatic applications that help the scientific community to unravel the complex relations between the consumption of phytochemicals and their expected effects on health.