Trial and error, arguably the most common approach to problem solving, is frequently delegated. For example, pharmaceutical companies outsource aspects of drug development to contract research organizations, hedge fund executives hire researchers to mine data for profitable trading strategies, and governments regularly task consultants with finding creative solutions to novel problems. But how should a delegator structure a problem solver’s contract when the problem solver can choose to withhold the solution? We prove that the contract which maximizes the delegator’s payoff uses a mixture of ‘carrots’ and ‘sticks’ to incentivize the problem solver to conduct trial and error and report the solution in a timely fashion. At a high level, the delegator’s preferred contract (i) shares the (expected) profit generated by the solution with the problem solver in the early stages and (ii) effectively slows down (or terminates) the trial and error process in later periods.