I guess that may work a little better, but I'm still more thinking about fail over if the API returns an error.