The Company was considering marketing a new product in a section of the highly competitive food market. Consumer tests on samples of product from a small pilot plant had given favourable results. The decision had then to be taken either to put the product on national sale as quickly as possible or to proceed more cautiously by test marketing in a limited region first, to get a better indication of the likely national sale. The former course involved risking large expenditures on plant and promotional activities; the latter approach allowed the collection of additional information before deciding whether to incur the major part of the expenditure, but increased the time before it would become an established profit-making product. In trying to relate their information to the marketing decision to be made, our market research colleagues felt that there must be a logical way to make the decision. Accordingly, it was decided to try to formulate the problem in terms of a mathematical model.