It obviously won't save the government £10m.
Like when you ask people how much they drink, then double it and double it again, any government figures about saving money on benefits should be halved and then halved again. And then the figure you end up with regarded as a wildly optimistic best case scenario.
Out of interest, counting reduction in benefits set against money paid out to workfare etc providers, increased costs at job centre, money paid to ATOS, are there any good estimates for how much has actually been saved so far?