The paper proposes an approach for a smart monitoring of the life-cycle of innovative graphene-based filters for water remediation in presence of pollutants. The measurement technique is based on suitable figures of merit that analyze the time variation of the electrical impedance frequency spectrum. The proposed study considers the remediation of two toxic industrial pollutants, such as the acetonitrile and the 2,4dichlorophenol. The contribution of the paper is twofold. The first is the demonstration of a reliable monitoring set-up that is able, for the selected use cases, to correlate in real-time the behavior of the electrical impedance of the filter to its status, defined as "absence of pollutants" and/or "saturation". The second contribution is the proposal of suitable figures of merit, based on measurement of the impedance frequency spectrum, able to increase the measurement sensitivity and the reliability and to mitigate some sources of uncertainty typically associated to these kinds of set-ups and measurements.Results show that the proposed graphene-based filters combine very good filtering capability and high sensitivity of the electrical impedance to the considered pollutants. These results suggest further investigations with other pollutants and the potential use of this technique for the predictive maintenance of the water filters in industrial applications, by endowing the graphene filters of smart sensing devices.