In accordance with a 'people centred' vision, this paper critically examines current approaches to smart cities benchmarking. In particular, by means of correlation analysis and Principal Component Analysis (PCA) we put in evidence present limitations of city rankings and, as well, the emergence of different perspectives for data interpretations. To follow, a possible redesign of the 'Smart Cities Analytics' grounded on the traces left by individuals, is suggested. In particular, as an example, we focus on the potentiality offered by automatic text analysis to extract people perceptions and expectations that, in turns, demonstrate the need to integrate bottom-up and top-down approaches to city benchmarking. Finally a novel definition of smart city based on the territorial state of flow is proposed and, as a consequence, a novel path toward smart city benchmarking suggested.