Habitat restoration is one of the actions to reduce landscape fragmentation by promoting connectivity and thus biodiversity. But knowing where to implement these habitats is a major issue and planners lack guidance for answering this question, in particular when it involves multiple species and over a large area. We proposed to combine biological data, habitat suitability models and spatial graphs to improve multiscale and multispecies connectivity in Ile-de-France, a highly artificialized region of 12,000 km2. The framework consisted of i) modeling habitat suitability for eight pond-dwelling species (Alytes obstetricans, Bufo bufo, Epidalea calamita, Hyla arborea, Rana temporaria, Salamandra salamandra, Triturus cristatus, and Natrix natrix), ii) modeling the ecological network for each species, iii) prioritizing each sampling point depending on the gain in connectivity if a new pond was created there and iv) combining single-species results to identify the areas that could improve multispecies connectivity. The multivariate statistical analysis revealed that transitional forest environments appeared to be the most strategic areas for improving multispecific connectivity (at least for 5 species). Targeted addition of habitat within an ecological network can increase habitat density in deficient areas and reconnect network sub-parts. This approach is therefore promising to guide conservation actions and “no net loss” biodiversity measures, especially the final stage of offset in the mitigation hierarchy.