Especially if the "right" model in the end isn't really compatible with the first, wrong one and you can't afford to mess up the data.
Especially if the "right" model in the end isn't really compatible with the first, wrong one and you can't afford to mess up the data.