The value of the initial guess is insignificant to whether the process convergences on a final answer.

If the gradient to the curve at the initial guess is f’(x) = 0, then we cannot use the Newton-Raphson method.

When the value of f(x) is "close enough" to zero, the iteration process is stopped and the resulting value of x is the estimate of the root.