The primal feasible point gives you a bound for the optimum value of the dual. Since you have a quasi-optimal primal point, and there is no duality gap, the primal function value at this point is also a good approximation for the optimum value of the dual problem.
But I can see how this helps you to find a quasi-optimal solution to the dual. I guess it doesn't.