The promise of any small improvement in the performance of light-harvesting devices, is sufficient to drive enormous experimental efforts. However these efforts are almost exclusively focused on enhancing the power conversion efficiency with specific material properties and harvesting layers thickness, without exploiting the correlations present in sunlight -- in part because such correlations are assumed to have negligible effect. Here we show, by contrast, that these spatio-temporal correlations are sufficiently relevant that the use of specific detector geometries would significantly improve the performance of harvesting devices. The resulting increase in the absorption efficiency, as the primary step of energy conversion, may also act as a potential driving mechanism for artificial photosynthetic systems. Our analysis presents design guidelines for optimal detector geometries with realistic incident intensities based on current technological capabilities