Customer data cleanup is safest when the app makes the decision visible and the merchant keeps responsibility for the final action. Customer Data Health is built around that sequence.
Start with the store boundary
Open the app from the intended Shopify store and confirm the store context before starting a scan. Do not paste customer exports into support or use a different store to test a destructive action.
Treat the scan as a queue for review
A finding is a signal, not a verdict. Read the explanation, compare the masked values and check whether the proposed action fits the merchant's own knowledge of the customer relationship. Similar names or addresses do not prove identity.
Preview before approval
Open the before-and-after preview and confirm which record survives, which fields change and whether the action is reversible in the merchant's own process. If the preview is incomplete, leave the item in review.
Reconcile the result
After an approved action, check the audit history and the issue status. A failed or blocked action should stay visible as such. Do not treat an empty result as proof that the customer database is perfect; it means that the enabled detectors found no actionable result for that run.
Read the Customer Data Health documentation for the full workflow and the privacy policy for the data boundary.