eCommerce data cleanup: attributes, images and product content

Normalize catalogue data from multiple sources using consistent fields, safe formatting and reviewable changes.

Suppliers may use different names for the same material, unit or size. Create a shared attribute model and mapping rules while retaining original values for review. Rules must not merge different products merely because their titles look similar. Normalization should improve filters and search without changing commercial identities. Ambiguous values enter a review list, allowing the team to decide whether they represent synonyms, different specifications or genuinely missing information.

Remove unnecessary formatting, unwanted inline styles and unsupported HTML from imported descriptions. Text then follows the store’s typography. Validate links and images under appropriate rules and usage permissions. Cleanup must not execute imported code or treat unknown content as trusted instructions. Define permitted markup explicitly so useful lists and emphasis remain readable while external formatting cannot break product pages or interfere with the surrounding interface.

Check image accessibility, correct product associations and required dimensions. Size optimization should not change the product’s appearance. Validate SKUs and EANs separately from descriptions. If a code is missing, create an internal identifier under an agreed policy rather than inventing a manufacturer barcode. Image-processing failures should leave approved assets intact and flag the affected records for inspection, avoiding empty galleries or misleading replacements on live product pages.

Each batch can preview changes before updating the catalogue. The team sees normalized fields, missing data and products needing attention. Original values and new versions remain distinct. Protect manual editorial work and test new rules on a small sample before bulk application. Keeping the rule version alongside the change report makes later troubleshooting easier, particularly when the same supplier changes field names or delivers different units in a new export.

Choose examples with Greek text, mixed units, unusual HTML, duplicate identifiers and broken images. Repeating cleanup should not repeatedly alter an already correct field. Check filters, search, feeds and page appearance after changes. Success means a consistent catalogue that can be maintained, with visible exceptions rather than hidden errors. Include a reversal procedure for incorrect bulk mappings so the team can recover original values without reconstructing the entire catalogue.

For a furniture store, one supplier provides lengths in centimetres and another in millimetres. Normalize both to the agreed unit, retain original values and flag implausible dimensions. Width filters then use consistent numbers without changing product identifiers or edited descriptions. Test the conversion against known items before extending it to every supplier record.