The survival of a business is based on how to predict and satisfy customer's needs. In general, the precision of demand forecasting is closely related to budget invested in forecasting and the nature of demand. In this paper, we develop a single-product single-period model to study how to determine the appropriate amount of budget spent in forecasting. We find there is an upper bound of the budget and is denoted as expected value of perfect information(EVPI). Finally, we discuss the relation between EVPI and model parameters in the sensitivity analysis.