I didn't see budget up at the top.
Tabletop exercises with surrogates are fine and dandy, but every so often you need to actually implement the critical pieces if at all possible. Full HA systems make this easier to do, but a failover, test, failback is insufficient; if you believe you can actually run the business on your alternate site, then do so for a day or a week.
Some situations a good test run for a day or a week will find (i.e. communicate using cellphones... except that half the cellphones end up with dead batteries, which a real test will show... and your failover IO capability is grossly insufficient to do what you think it would), and some you only see in particular real disasters (cell phone towers dying, rendering the remaining cellphones expensive and short-lived flashlights).