Sample and verify
What the data cannot answer comes back as a question.
First the universe is defined, then it is measured, then it is built. Where two sources disagree, the conflict is recorded rather than resolved silently. The file ships with its own limitations written down.
Franchise handling
Separating independently owned sites from company-owned ones where the evidence allows.
Date normalisation
Putting every date in one format and one timezone before any comparison runs.
Second-opinion pass
Having somebody who did not build the file check it against the brief.