This paper describes a study in Smart Grid Tariff Design. We use historic Smart Meter electricity usage data and Representative Load Profiles (RLPs) to explore Time-of-Use (ToU) tariffs. We consider the problem from the perspective of a supplier who wishes to set tariffs to maximise revenue, and has to anticipate responses from consumers who wish to minimise their costs. This is a typical bilevel optimisation problem, but with additional challenges as consumers have uncertain electricity usage. We create test RLP instances using historical Smart Meter data and demonstrate the use of a robust optimisation framework to address uncertain usage patterns. Our modelling and optimisation approach offers a framework for suppliers, system operators and innovators to explore and evaluate potential ToU market offerings.