The future of biodiversity and ecosystem services hinges on harmonizing agricultural production and conservation, yet there is no planning algorithm for predicting the efficacy of conservation investments in farmland. We present a conservation planning framework for countryside (working agricultural landscapes) that calculates the production and conservation benefits to the current baseline of incremental investments. Our framework is analogous to the use of reserve design algorithms. Unlike much countryside modeling, our framework is designed for application in data-limited contexts, which are prevalent. We apply our framework to quantify the payoff for Costa Rican birds of changing farm plot and border vegetation. We show that installing windbreaks of native vegetation enhances both bird diversity and farm income, especially when complementing certain crop types. We make predictions that differ from those of approaches currently applied to agri-environment planning,: e.g., although habitat with trees has lower local species richness than farm plot habitats (1-44% lower), replacing any plot habitat with trees should boost regional richness considerably. Our planning framework reveals the small, targeted changes on farms that can make big differences for biodiversity.