Global products of phytoplankton functional types (PFTs) derived from multi-sensor ocean color (OC) data provide important long-term biogeochemical quantifications indexed by chlorophyll a concentration (Chl a) of PFTs, including diatoms, haptophytes, prokaryotic phytoplankton, dinoflagellates, and green algae. Due to the distinctive lifespans and radiometric characteristics of ocean color sensors, the consistency of the PFT products derived from different sensors needs to be assured to establish a complete and systematic frame for long-term monitoring of multiple PFTs on a global scale. This study introduces a machine-learning-based (ML-based) correction scheme to eliminate the discrepancies between different sensors' PFT products. The correction scheme is applied to the Sentinel 3A/B Ocean and Land Colour Instrument (OLCI)-derived PFT data to match them with the PFT data derived from GlobColour-merged ocean color products using the overlapped period. This correction has generated consistent PFT data across the sensors, enabling the analyses of multi-year PFT observations by describing their variability and 2-decade trends. Analysis of PFT time series has revealed an increasing trend in diatoms and dinoflagellates and a decreasing trend in haptophytes and prokaryotic phytoplankton on a global scale. The overall trend in green algae remains relatively stable, although with some spatial variations. These PFT trends are more significant in high latitudes and coastal regions (and also in the equatorial region for prokaryotic phytoplankton). The anomaly of PFTs in 2023 shows significant increases in Chl a of diatoms and dinoflagellates (+24 % and +9.4 %, respectively) but only weak changes in Chl a for prokaryotic phytoplankton (-2.1 %) and haptophytes (∼ 1.6 %). These consistent time series data will act as an important ocean indicator to infer possible changes in the marine environment.