This is really the top priority: determine that you are building something like the right product.

In many cases, you don't even need to build much/any product to figure this out sufficiently.