Lake restoration is a difficult task, where complexity, uncertainty and lag times of catchment responses interact with long residence times, regime shifts and feedbacks from internal (bottom-sediment) nutrient loading in lakes. Given these complexities, it is critical to optimise the efficiency and effectiveness of monitoring programs for detecting water quality improvements from mitigation actions. As part of an Our Land and Water National Science Challenge research program (in Aotearoa, New Zealand), we aimed to develop an optimised stream and groundwater monitoring network coupled with a targeted lake monitoring program. The outcome is a catchment-scale monitoring network that captures the hydrological connectivity and interdependencies of surface water, groundwater and lakes. We use statistical and geospatial modelling to derive discharge and nutrient concentrations in inflows to lakes, and hydrological, morphological and semi-empirical equations to assess lake responses and design the optimal monitoring location, duration and frequency. Key lake monitoring variables include total nitrogen, total phosphorus, chlorophyll a and Secchi depth. Finally, we assess the costs of monitoring to provide an assessment of the usefulness of grab samples versus high-frequency sensors, as properly calibrated sensors have the potential to reduce the duration of monitoring required for statistically significant change detection. Our approach is applicant to assessments of lake monitoring at catchment, regional and national scale.