A randomized experiment run in partnership with a large agri-business sugar company in Kenya shows that a mobile-based query system can improve the company's performance in the management of the provision of inputs to the company’s cane suppliers and generate positive geographic spillovers. Given the limited sample size, it was not possible to confirm or to rule an effect on sugar cane yields. In another intervention, the company sent test messages with agricultural advice to cane farmers. In contrast to a previous trial, this treatment had no significant impact on cane yields. The research team is currently investigating the potential reasons for the different outcomes in response to the treatment.