In the current global business environment, it is very important how to allocate products
from the manufacturer to distributors for the efficient response to customer orders. Imaginary
demands ordered by distributors are the big obstacle to the efficient allocation of products. If
the manufacturer can become aware of imaginary demands included in distributors' orders, it
is possible to realize the high level order fulfillment through the efficient allocation of products
using neural network. We propose new allocation policy considering imaginary demands. In
this study, the backpropagation algorithm, one of algorithms in neural network, is used to find
imaginary demands from the distributors' orders.