Ecommerce Product Data Migration: Map Fields Before Importing the Catalogue

A catalogue migration can import every row and still produce a store customers cannot use. Variants may become separate products, filters may contain inconsistent values and a price may refer to the wrong pack size. The central task is preserving meaning across two different product models.

A field-mapping workbook makes those decisions visible before the first import. It should describe identity, relationships, transformations and validation, not just match column headings. This guide concentrates on product data rather than redirects, search rankings or the wider platform cutover.

Start with product identity and sellable variants

Identify which source record represents the parent product and which represents the item a customer can buy. A parent might describe a shirt, while each combination of size and colour has its own stock and price. Keep those relationships explicit in the destination.

Hypothetical example: a workwear catalogue contains one jacket in three sizes and two colours. It has six sellable variants, not six unrelated jackets. One discontinued combination remains in historical orders but should not become available merely because its source row exists.

Choose stable identifiers for matching imports to existing records. Names and descriptions are poor matching keys because editors change them. Inspect whether source SKUs are unique, whether leading zeroes matter and whether a supplier has reused a code for a different item.

Build a mapping with decisions attached

For each destination field, record the source, transformation, required status, allowed values and response to invalid data. Name the person who approves ambiguous mappings. Keep an example source value and expected destination value beside the rule.

Distinguish an intentionally empty value from a missing source value. An empty optional subtitle may be acceptable; an unknown fulfilment weight may require review. Avoid replacing every missing value with a convenient default simply to make the import complete.

Normalise attributes without erasing distinctions

Create an approved vocabulary for filters and options. “Navy”, “navy blue” and “Navy Blue” may belong together, but visually different finishes should not be collapsed because their names are similar. Ask a product specialist to approve the vocabulary.

Keep display labels separate from machine identifiers where the destination supports that distinction. This allows an editor to improve wording without changing the identity used by integrations. Preserve meaningful units for dimensions, capacity and weight instead of embedding them inconsistently in free text.

For category mapping, describe the browsing purpose. A supplier's purchasing hierarchy may not suit customers searching the new store. Allow products to appear in relevant customer categories while preserving the source classification separately if operations still need it.

Check the commercial meaning of prices

Record currency, price basis, quantity breaks and whether a field represents a regular or promotional amount. Verify the store's agreed tax display configuration with the responsible business owner. Do not infer the meaning of a numeric column from its heading alone.

In the hypothetical workwear example, a price for a carton must not become the price for one jacket. Test the product page, basket and order summary with the same variant and quantity. This exposes mistakes that a spreadsheet comparison may miss.

Schedule-dependent promotions need their own rules. Decide how expired offers, future start dates and missing end dates translate. Importing an old promotional flag without its timing can create an unintended live offer.

Treat media and descriptions as structured data

Inventory image files, alternate text, display order and variant associations. Check that file references resolve in the destination and that the featured image matches the selected option. Detect reused filenames rather than assuming identical names identify identical assets.

Inspect description formatting for unsupported markup, embedded links and copied layout fragments. Preserve useful headings and lists while removing obsolete presentation code through an agreed transformation. Review representative long descriptions, not only short products with clean text.

Attach documents such as size guides or installation instructions to the correct product family. Confirm that users can open them and that obsolete versions are excluded. A successful media upload does not prove the document belongs to the item displaying it.

Rehearse, reconcile and control reimports

Choose a sample covering ordinary products, many variants, missing fields, bundles and discontinued records. Import it into a test catalogue and have merchandising staff inspect browsing, filtering, selection and basket behaviour. Record exceptions against the mapping rule that needs correction.

Compare parent and variant counts separately, then check relationships and selected field values. Keep a rejection report with source identifiers and reasons. An import that quietly skips troublesome rows is not ready for approval.

Define what a repeat import updates and what it preserves. If editors correct destination descriptions during migration, a later run should not silently erase their work. Decide how source changes after the first extraction are captured and reconciled before release.

Stock deserves a separate ownership model; see stock synchronisation and reservation rules. For catalogue implementation, review Visuosofts' website development service and contact the team with a sanitised sample export, variant examples and unresolved mapping questions.