Lodovico Capuano, Massimo Vaglio, Rohit S. Chandramouli, Chantal L Pitte, Adrien Kuntz, Enrico Barausse

Inaccurate modeling of gravitational-wave signals can introduce systematic biases in the inferred source parameters. As detector sensitivities improve and signals become louder, mitigating such waveform-induced systematics becomes increasingly important. In this work, we assess the systematic biases introduced by an incomplete description of the ringdown signal from massive black hole binaries in the LISA band. Specifically, we investigate the impact of mode truncation in the ringdown template. Using a reference waveform composed of 13 modes, we establish a mode hierarchy and determine the minimum number of modes required to avoid parameter biases across a wide range of LISA sources. For typical systems with masses $\sim 10^6$--$10^7\,M_\odot$ at redshifts $z \sim 2$--$6$, we find that at least 3--6 modes are needed for accurate parameter estimation, while high-SNR events may need at least 10 modes. Our results are a window-insensitive lower bound on the minimum number of modes, as more modes may be needed depending on the choice of time-domain windowing of the post-merger signal.