Low and decreasing biodiversity in agricultural landscapes threatens food production in the long run due to a lack of ecosystem service provision. From an ecological perspective, mitigating policy measures such as flower strips are often mislocated due to farmers’ various economic considerations. High opportunity or conversion costs, induced by a soil’s quality or the land’s value, possibly inhibit farmers to sow flower strips in the most suited regions, which impedes biodiversity improvements. To overcome possible inefficiencies in flower strip policies, the research identifies whether flower strips are typically sown in already biodiverse areas. Further, it examines whether spatial targeting of flower strips based on biodiversity, land value and soil quality according to their marginal benefit is relevant in North-Rhine Westphalia. Field-level IACS and other vegetation data are employed to calculate landscape metrics and to derive biodiversity indicators. In a correlation and linear regression analysis, the relations between flower strip abundance and area share, levels of landscape biodiversity, land value and soil quality are analyzed. This aims at identifying whether biodiversity gains brought about by flower strips coincide with levels of biodiversity, land value or soil quality. If so, this enables characteristic-based spatial targeting of flower strips to regions where they yield the highest benefits, albeit economic obstacles. It is found that neither flower strip abundance nor their area share depend on the biodiversity level of a landscape. However, flower strip abundance and area share are drivers for biodiversity improvements. As flower strip-induced biodiversity change in a landscape is not influenced by the regional soil quality or land value, spatial targeting based on only these two factors is not relevant for North-Rhine Westphalia. The approach is transferrable to any region with agricultural plot-level and vegetation data and allows consideration of other influential factors than the ones considered here, as long as they are georeferenced.