Background: Achieving the goals of the EU Biodiversity Strategy for 2030 requires reliable, evidence‑based tools to guide conservation planning. Habitat suitability models can help identify priority areas by predicting suitable habitats for taxa of conservation concern. However, their usefulness depends heavily on the quality of input data. Using heterogeneous data sources, especially citizen-science occurrences, poses challenges for ensuring quality-informed use of these records. In addition, inconsistencies in the selection of environmental variables can compromise the reliability of the results, particularly when combining predictions across multiple species. Therefore, transparent species selection and rigorous, harmonised data preparation are essential for producing robust spatial assessments that can effectively support national‑level conservation decisions. New information: We present conceptual framework and methodological rationale for a structured dataset for model-based conservation planning in Latvia, from species selection and observations to species-specific ecogeographical variable selection. The framework was developed and the dataset created as a part of the state research programmes project "HiQBioDiv: High-resolution quantification of biodiversity for conservation and management". The dataset provides a basis for multi-taxon habitat suitability modelling that can be further used in site prioritisation for conservation as well as in ecological and socio-economic assessments. As a part of this dataset, we provide 162,186 expert-filtered occurrences from 464 taxa, most of which were previously unavailable to the international community. Keywords: habitat suitability models, multispecies prioritisation, spatial planning.