Keep the original
Retain an untouched source copy. Record its filename, export time, source system, and record count. Work on a separate copy and note each transformation.
Profile the file
Free CSV import checklist
Check the file. Then make the move.
A reusable checklist for moving CSV records into a CRM, database, or business application. Preserve the original values, test the mapping, and account for what changed after the import.
- Included
- 16 checks with pass conditions and blank status and notes fields.
- Access
- Free download. No account or email required.
- Reuse
- Adapt it for your own or your organization's internal work, including commercial use.
Before importing
Eight decisions to make before the file goes live.
Use the download to record Pass, Fail, or N/A with evidence in Notes. Resolve failures before approval. The tools help inspect a file; your destination's import rules determine whether it is ready.
Check the CSV structure
Confirm the destination accepts the encoding and delimiter. Inspect accented text, embedded line breaks, and quoted commas. Count parsed records, since a record can span several text lines.
See a test caseMap every column
Document the source column, destination field, expected type, required status, and blank-value behavior. Review extra, missing, or repeated headers before mapping.
Inspect missing valuesPreserve identifiers
Treat identifiers as text. Confirm the matching key is complete and unique for this import. Decide whether the operation creates, updates, or upserts records before using a key to match them.
Check key candidatesMake conversions explicit
Keep raw amounts and dates as text during review. Specify currency, decimal separator, date order, and time zone where relevant; preview any approved conversion before applying it.
Review date formatsReview duplicate candidates
Check exact duplicate keys separately from similar names or records. Confirm whether each candidate is a repeat or a distinct record. Record the decision before merging or deleting anything.
Find near duplicatesRun a representative test
Use a sandbox or documented preview mode. Include blank values, quoted text, long IDs, duplicate keys, and ambiguous dates. Verify rejection behavior and a recovery method before approval.
Review the go-live checksReconcile the result
Save the import log and post-import export. Explain created, updated, skipped, and rejected records under the destination's rules. Compare matched records by key and inspect unexpected field changes.
Compare CSV exportsFictional test case
A small sample can reveal a large problem.
These invented values are test cases, not customer data. Include comparable edge cases from your own import contract in a small test file.
Leading zeros are part of the ID
001 and 1 are different text identifiers. Opening and saving a file with automatic numeric conversion can remove that distinction. Set the identifier field to text and confirm it survives the test import.
A comma can belong to a value
In a comma-delimited CSV, "Acme, Test" is one quoted field. Verify that the importer keeps the company name together instead of shifting the remaining values into the wrong columns.
A familiar date can be ambiguous
03/04/2026 does not establish whether March or April is intended. Confirm the source's date convention, document the conversion, and verify the resulting date in the destination.
Go-live review
Know how you will verify and recover.
A successful upload is only one part of the check. Decide what success means before you modify production records.
Record approval
Name the reviewer and import owner. Record the target environment, file version, mapping, create/update rules, test findings, and accepted exceptions. Make it clear which exact file and settings were approved.
Plan a usable rollback
Export the affected records and retain the identifiers needed to find new records. Check whether the system supports reversing this operation, restoring field values, or recovering a backup. Confirm the recovery method in the test environment.
Explain the counts
Check how the importer defines each outcome; some categories overlap. For exclusive outcomes, accepted plus rejected plus skipped should account for the submitted records. With updates, the destination's total record count may stay the same.
Keep a record
Use the checklist on the next file, too.
The CSV has five columns: step, check, pass_condition, status, and notes. The Markdown version includes a place for the file, mapping, owner, approval, and recovery details. Adapt the checks to your destination.
Keep a repeatable comparison workflow.
Data Workbench compares two CSV exports by the identifiers and fields you choose. The offline Browser Workbench includes saved settings and CSV/JSON reports; the Reconciliation Kit provides a JavaScript engine and command-line runner. Each edition is a separate $49 one-time purchase.