We use an economic experiment to identify how people design a mechanism. Our experimental framework is adapted from the market framework utilised by Maskin and Riley, who considered a monopolist seller of a homogeneous good that uses non‐linear pricing under incomplete information. Approximately 30 per cent of participants’ offers in each treatment were incentive‐compatible (separating) double offers, a requirement of the optimal mechanism. A similar percentage of offers in each treatment were single (non‐separating) offers, which do not require participants to address incentive compatibility. After sufficient rounds, the design of double offers converged to the schedule identified by Maskin and Riley, while the design of single offers converged to the optimal non‐separating (single) offer.