Ocean primary production is a key process that regulates marine ecosystems and the global climate, but its estimation is still affected by multiple uncertainties. Typically, the chlorophyll-a concentration (CHL) is used to characterise this process, as it is considered a proxy of phytoplankton biomass. To date, the most common observing systems for studying CHL are ocean colour satellites and BGC-Argo floats. Those are complementary systems: satellite observations provide global coverage but are limited to the ocean surface, while BGC-Argo floats provide punctual observations along the whole water column. Comparison of these two observing systems has been performed only at regional or single-float scales, while at global scale this results in large uncertainties due to the relatively low and irregular BGC-Argo coverage. Here, we propose a different method, by comparing satellite and BGC-Argo climatological annual time series within seven different bioregions, each characterised by a homogeneous phytoplankton phenology, allowing us to smooth the uncertainties. By comparing the mean values, the amplitudes, and the shapes of the two time series, we are able to identify regions (a) where they agree (58-61% of the ocean surface area); (b) where the BGC-Argo float network should be extended (generally regions with less than 5 profiles each 100x100 km2 square); (c) where the discrepancy is likely due to satellite or (d) BGC-Argo performance. Use of either BGC-Argo and satellite data in regions b—d should be carried carefully and we provide, for each region, suggestions on which system could be affected by the largest uncertainties.