My first startup made that mistake. We built the entire product before getting any customers. Since then, I've always started new projects as pseudo-consulting gigs. That way, the product is designed around a real need.