In absence of the real prototype or sample batch, how could we estimate demand?

I can think of:

1. Surveys. Explain the product to the user. Ask him would it be helpful? Ask him how much he would pay for it?

The accuracy of the estimate would depend largely on the sample size, and a bigger sample size of users would tend to give more accurate results. To get a bigger sample size survey is expensive. Sometimes as expensive as making a prototype itself.

2. Industry experts. Ask them and you'll probably know!

They are no less expensive than the survey itself.

All else I have read about are derivatives of these two in one way or the other with some calculations. I can't think a way of estimating demand out of these. Are there any?