Integrated photonic sensors are promising devices to detect the presence of biological and chemical substances. Especially, silicon photonics features scalable, compact, and robust applications in, inter alia, the medical, pharmaceutical, and chemical industries for decentralized, real-time sensing. In an effort to continuously lower the limit of detection and build ultra precise devices, their sensitivity is steadily increased. The optical loss, however, is often neglected. This reduces the available intensity in the transmission spectrum and, thus, restricts the achievable limit of detection. We optimize a bimodal waveguide interferometer for environmental sensing based on a silicon nitride platform. Thereby, we combine considerations regarding the sensitivity, optical loss, and coupling into a figure of merit to design the single-mode and the bimodal platforms. To that end, we utilize a recently developed model of surface-roughness-induced scattering to include the estimated propagation loss in the design process. This model uses the simulated modes and measured autocovariance in the surface roughness for an estimation of the propagation loss in the real platform. We observe that considering the optical loss favors platforms of medium height for a quasi-transverse-electric polarization. In consequence, we demonstrate drastic changes in the design choices compared to pure sensitivity-focused waveguide platforms. Thus, we show that a holistic design process is essential to fully utilize the potential of integrated photonic devices. Ultimately, we are convinced that this methodology will be beneficial in making integrated photonic sensors more sensitive.