Besides the transient effect, the passage of a gravitational wave also causes a persistent displacement in the relative position of an interferometer's test masses through the nonlinear memory effect. This effect is generated by the gravitational backreaction of the waves themselves, and encodes additional information about the source. In this work, we explore the implications of using this information for the parameter estimation of massive binary black holes with LISA. Based on a Fisher analysis, our results show that the memory can help to reduce the degeneracy between the luminosity distance and the inclination for binaries observed only for a short time (∼ few hours) before merger. To assess how many such short signals will be detected, we utilized state-of-the-art predictions for the population of massive black hole binaries and models for the gaps expected in the LISA data. We forecast from tens to few hundreds of binaries with observable memory, but only ∼ O(0.1) events in 4 years for which the memory helps to reduce the degeneracy between distance and inclination. Based on this, we conclude that the new information from the non-linear memory, while promising for testing general relativity in the strong field regime, has probably a limited impact on further constraining the uncertainty on massive black hole binary parameters with LISA.1 Nevertheless, in the ∼ 100 GW signals observed to date there is only limited evidence for higher multipole content (with no evidence at all beyond = 3) [36] and only one measurement of strong-field precession has been claimed [37] (though some doubt has been cast on this claim due to data-quality issues [38,39]).