Economic data arrives with theatre attached: flash estimates, instant reactions, a market move before lunch. The revision arrives later, usually quieter, and often closer to the truth.
That asymmetry matters. If a preliminary jobs number is overstated, the first story is remembered as strength and the revision is filed as housekeeping. The public record changes, but the public impression does not.1
The honest habit is simple: treat first estimates as provisional and check whether the series has a revision history. A number that is designed to be revised should not be reported as if it were carved in stone.