Off the top of my head
It needs to pay for itself on a timescale less than the period for which you have a reasonable hope of predicting the price of electricity.
Deployment will be limited to those properties that can retrofit the battery and that itself may be costly if the battery is large. New builds would be fine, though, and that may be a large enough market for starters.
On the plus side, if this can be deployed widely then it is a game-changer for electricity generation because you wouldn't need power stations that can switch on and off at short notice.
So, um, what breakthrough in battery technology makes this possible in 2015?