How to mask Excel and CSV data for testing
Developers and analysts often need realistic table relationships without unnecessary personal details. CloakData prepares masked working copies of structured files locally in the CloakDoc Windows app, with Pro or Enterprise.
Choose what each column should retain
Keep fields needed for the test, tokenize identifiers that must remain consistent, and redact fields that are unnecessary. Review every kept column. Names and contact details can also appear in comments or free-text descriptions inside cells.
Preserve relationships between tables
Suppose a customers CSV has a customer_id column and an orders CSV has a buyer_id column. Assign both to the same relationship group. With the same workspace key and exact source value, matching values receive matching tokens so the exported tables can still be joined.
Matching is case-sensitive: leading zeros matter, and 001 differs from 1. Existing duplicate primary keys or missing foreign keys are not repaired. Tokens are text and do not preserve the original value's type or length; adjust your test schema accordingly.
Prepare an Excel workbook
Resolve formulas and merged cells before import. Review hidden worksheets as well as visible sheets. Export creates a new values-only workbook with text cells; original styling, comments and metadata are not copied. Excel dates remain stored serial values, so check how your downstream application interprets them.
Run a small sample before a full export
- Import representative CSV or Excel files into CloakData on Windows.
- Review the tables and assign keep, tokenize or redact actions.
- Group related key columns and protect the secret workspace key.
- Process and export transformed copies.
- Check row counts, retained fields, join behavior and downstream import compatibility.
CSV inputs use UTF-8 with unique, nonempty column headers. Only CSV, XLS and XLSX files are accepted. XLS imports are limited to 32 MiB; Excel exports are values-only XLSX workbooks.
Masking is not synthetic data generation
Tokenization replaces selected values consistently; it does not generate realistic fake names, valid replacement bank accounts or statistical synthetic records. Pseudonymized data can remain identifiable through unchanged fields or linkage. Use a fully synthetic dataset if the test does not require real source records.
Keep future runs consistent
Save your workspace key securely if future exports must match. It is not uploaded or automatically persisted. Both paid plans support up to 1,000,000 records per run across selected tables, subject to practical device capacity. Review local working files and exported copies when the task is complete.