The inhibition of the production and distribution of bottled water has become a serious problem in the survival of the community and the company, so there is a need for a solution to this problem both short-term and long-term solutions. One of the things that can be done by the company or management is that the right amount of production and distribution is by forecasting sales. Sales forecasting is the process of predicting which products will be sold in the future made based on data that has ever happened. This paper aims to determine the level of accuracy of the use of Backpropagation ANN in estimating the sales of bottled water.The ANN architecture used is 12-10-1 with the MSE value of 0,00043743 and the MAPE value of 6.88%. Forecasting sales results of Robong Holo 600ml brand using Backpropagation ANN for 2019 is 2271 pcs in January, 2019 pcs in February, 1358 pcs in March, 917 pcs in April, 462 pcs in May, 324 pcs in June, 739 pcs in July, 370 pcs in August, 367 pcs in September, 1073 pcs in October, 765 pcs in November and 1388 pcs in December.
Keywords— AMDK,Backpropagation,Jaringan Syaraf Tiruan,Penjualan,Peramalan