THE SHORT ANSWER
Audit a mapping by following representative values from source to destination. Check field meaning, data type, units, blank handling and identity preservation. Save the transformation and expected result so another operator can explain the output without reconstructing the original implementation.
Document meaning before column names
Two columns with similar names may represent different facts. A source price might include tax, use minor units or apply only to one market. A group ID may describe a supplier collection rather than a variant family. Ask what each value means in the selling system.
Create a mapping record with source field, destination field, transformation, fallback and owner. Mark unresolved assumptions explicitly. An empty owner or an unexplained constant is a useful audit finding, even if the export currently passes a basic parser.
Review transformations and empty values
Inspect trimming, case changes, numeric conversion, currency formatting and text assembly. Pay particular attention to identifiers that look numeric. A conversion that removes leading zeros can break identity while leaving every row syntactically valid.
Define null, empty and zero separately. Decide when a fallback is appropriate and when missing information must remain an exception. Avoid silently replacing an absent price, material or identifier with a plausible-looking value that nobody supplied.
Build a representative mapping sample
Include a simple product, a multi-option variant, a discounted product, an unavailable item and a record with optional fields missing. Add source-specific edge cases such as accented text, long model references and embedded delimiters.
For each sample, show the original values and expected output. Review the result against the destination’s current contract, not just yesterday’s file. A working historical mapping can become inappropriate when a source or receiving interface changes.
Turn findings into controlled revisions
Prioritize identity and commercial-fact issues before cosmetic formatting. Preview the affected scope of a rule change and compare it with the last usable release. Keep mapping revisions attached to release records so a regression can be traced quickly.
After correcting the mapping, rerun the sample and inspect a broader output slice where necessary. Then monitor the next scheduled import. A one-off manual edit does not fix a recurring transformation, and the next run may reintroduce the original problem.
ILLUSTRATIVE WORKED EXAMPLE
Make it concrete.
Illustrative example: a source price is stored as 3499 minor units. The audit finds that the export emits 3499 USD instead of 34.99 USD. The team corrects the documented unit conversion, checks several prices and verifies that the next recurring run uses the revised mapping.
Your practical checklist
- Record each field’s meaning, type and owner.
- Review units, fallbacks and blank behavior.
- Preserve identifiers as text.
- Save expected output for representative source cases.
- Attach mapping revisions to releases and retest scheduled runs.
Common questions
Can matching column names automate the whole audit?
They can suggest a mapping, but they do not prove semantic equivalence, correct units or appropriate source ownership.
How many sample records are enough?
Choose cases that exercise your actual source and transformation risks. A small deliberate sample is more informative than many identical ordinary rows.
Sources & scope
Official references checked 29 October 2026. Examples are illustrative, not customer results. Platform requirements can change; use current destination documentation for implementation decisions.
Our editorial method & corrections ↗