The brief sets the boundaries; everything downstream inherits them. Categories are checked against the underlying description, not the label attached upstream. Format is tested on a sample before the full run is delivered.
Outlier review
Looking at the extremes of each field, because that is where bad data announces itself.
Column reconciliation
Bringing differently named fields from separate feeds onto one agreed set of headings.
Supply assessment
Saying plainly where the available data stops, including when it stops short of the brief.
Send the awkward criterion first; it tells us faster whether this is possible.