Checklist
Data Cleanup Before Sharing Checklist
A final review for data files before they leave your device.
Tick items as you work through them. Nothing is saved — reloading or reopening this page starts a fresh copy. Print it if you need to keep your place.
Define the copy you are sharing
Write down who should receive the file and what they need to do with it. That boundary makes unnecessary columns and records easier to identify.
Make a separate sharing copy so cleanup cannot destroy the source data or remove fields needed for another use.
Check whether the recipient expects plain text, CSV, JSON, or another format before changing delimiters, ordering, or structure.
Keep a simple before count so deduplication and filtering can be reconciled instead of silently dropping more records than intended.
Remove sensitive values
Look for credentials, API keys, access tokens, private URLs, and connection strings. Revoke any real secret that was exposed, even if you remove it from this copy.
Delete names, addresses, contact details, identifiers, health information, and financial data unless the stated purpose genuinely requires them.
Notes and comments can contain sensitive information that a column-name review misses, so sample and search their actual values.
A clean data body can still travel with an identifying filename, author field, path, revision note, or other metadata outside the visible rows.
Check structure and consistency
Use the JSON Formatter and Validator for JSON data, then read the result. Valid syntax does not prove that field names or values are correct.
Remove accidental empty records and confirm that line breaks will be interpreted consistently by the recipient's software.
Case can be meaningful in identifiers, codes, paths, and JSON values. Normalize it only where the data contract says case is insignificant.
Confirm whether duplicate-looking lines are genuinely redundant. Two records can share a name while representing different people, dates, or transactions.
Compare the cleaned count with the original and account for every removal, filter, and deduplication step.
How to use this checklist
Sharing a file is not reversible. Once a copy has left, you cannot decide later what the recipient kept, forwarded, or indexed — which is why the work belongs before the send rather than after it, and why “I removed it afterwards” is not a remedy. A leaked credential is the clearest case: it stays usable until it is revoked, so deleting it from your copy changes nothing.
Two failure modes account for most of the damage, and neither is visible in the rows you are reading. The first is information carried outside the data — filenames, author fields, revision notes, and the path the file was saved from. The second is the share itself: a link scoped to “anyone with the link” rather than to a person, or an address completed by autocomplete to the wrong recipient. A file can be perfectly clean and still be disclosed to the wrong audience.
This checklist reduces avoidable mistakes; it is not a guarantee that data is anonymous, secure, or legally appropriate to disclose. That distinction is substantive rather than a disclaimer: removing names and obvious identifiers does not by itself make a dataset anonymous, because remaining fields can still single someone out in combination. NIST’s De-Identifying Government Datasets (SP 800-188) sets out the techniques and the residual re-identification risk that survives them. Rules also vary with the data, organization, recipient, and jurisdiction. Follow the requirements that apply to your situation.
The JSON Formatter and Validator processes JSON in your browser and can catch syntax errors. The List Sorter and Deduplicator can help inspect line-based lists. Neither tool decides whether information is sensitive or whether two records are truly duplicates—that review remains yours.