A quick glance at agricultural input use data from developing countries reveals a large dispersion in take up of improved inputs and practices across farms. One explanation is that this is a problem resulting from limited information, credit constraints, risk, poor input quality, and/or behavioral biases. An alternative view is that farmers are in fact making optimal adoption decisions, and differences among farmers instead reflect heterogeneity in a fixed factor such as soil quality. The authors test whether rainfed farmers in Tlaxcala, Mexico adopt tailored recommendations based on soil analyses and whether productivity improves as a result. The authors vary the level of information specificity (whether recommendations are based on the farmer's own plot or on a larger geographical area) because individually tailored information may be more effective but is also more expensive. The authors also vary whether farmers can choose what inputs to purchase by offering an inflexible grant that subsidizes only the recommended inputs or a flexible one that gives farmers the choice of which inputs to purchase.