Duplicate SKUs. Mismatched attributes. Broken taxonomy. Missing descriptions. These are invisible until they cost you search rankings, filter navigation and margin. MercuryMinds diagnoses and systematically fixes catalog data quality — at the SKU count that makes manual cleanup impossible.
Catalog Cleaning Services
Every catalog cleaning engagement starts with a data audit that quantifies the problem by type. The cleanup work addresses each problem category systematically — not randomly — so the output is a documented, measurably cleaner catalog.
Deduplication
Identifies all duplicate and near-duplicate product records — including cross-supplier duplicates with different titles — and merges them under a single canonical record with the most complete attribute set. Merge map documented and reviewed before any records are deleted.
Taxonomy
Reclassifies products that have been incorrectly or inconsistently categorised — fixing the taxonomy drift that accumulates when different team members assign categories differently over time. Category tree reviewed and agreed before reclassification begins.
Attributes
Normalises attribute values that have been entered inconsistently — "Red", "red", "RED", "Crimson" all become a single standard value. Removes attribute data that belongs in a different field. Fills missing required attributes from source materials where available.
Data Quality
Scores every SKU against a quality rubric — description completeness, attribute coverage, taxonomy accuracy, image count — and produces a prioritised remediation list. Highest-impact fixes addressed first rather than working through the catalogue sequentially.
Migration Debt
Cleans the data damage caused by a platform migration — encoding artefacts, lost formatting, broken variant relationships, incorrect field mappings and taxonomy that didn't transfer correctly. The cleanup that should have been part of the migration project.
Ongoing
Post-cleanup monitoring pipeline that flags new data quality issues as they are introduced — duplicate detection on incoming supplier feeds, attribute completeness checking on new SKUs and taxonomy validation before products go live. Catalog quality maintained, not just fixed once.
How It Works
No two catalogs have the same data problems. The audit comes first so the cleaning work addresses the real issues in the right order.
Full catalog export and analysis — duplicate density, attribute completeness, taxonomy consistency, description coverage, image count. Problems quantified by type and severity before any cleaning begins.
Prioritised cleanup plan produced from the audit — which problems to fix first based on business impact. Merge maps, reclassification rules and attribute standardisation logic agreed and signed off before work begins.
Cleanup operations run against a sample first — 10% of affected records reviewed before full run. Output validated against the quality rubric established in the audit. Full run executed only after sample validation passes.
Clean catalog delivered in your platform's import format with a full audit report — before and after quality scores, records merged, attributes standardised, categories corrected. Everything documented so the changes are reversible.
Catalog data cleaning is the process of identifying and fixing data quality problems in an ecommerce product catalogue — duplicate SKU records, inconsistent attribute values, incorrect taxonomy placement, missing required fields and encoding artefacts from data migrations. Unlike catalog enrichment (which adds new data), cleaning fixes data that already exists but is wrong, inconsistent or duplicated. MercuryMinds starts every catalog cleaning engagement with a full data audit that quantifies the problems by type and severity before any cleanup work begins.
Duplicate detection uses a combination of exact match (identical SKU codes, EAN/UPC codes or product titles) and fuzzy match (similar titles, matching attribute sets with minor variations) to identify all duplicate records across the catalogue. Each duplicate cluster is reviewed to identify the most complete record — the one with the most attributes, descriptions and images — which becomes the canonical record. Duplicates are merged into the canonical record before being removed, preserving the best available data from all versions. The merge map is documented and reviewed before any records are deleted, so the process is reversible.
Product taxonomy standardisation is the process of ensuring every product is assigned to the correct category in a consistent category hierarchy. In most ecommerce catalogues, taxonomy drifts over time as different team members assign products to categories differently, new categories are added without reclassifying existing products, and platform migrations transfer products to incorrect category equivalents. Taxonomy standardisation first audits the current category assignment against a reference taxonomy, then reclassifies incorrectly assigned products systematically. The result is a catalogue where on-site search filters, faceted navigation and category browse produce correct results.
A catalog cleaning project for 10,000 SKUs typically takes 2–4 weeks depending on the severity and type of data problems identified in the initial audit. The audit itself takes 2–3 working days and produces a remediation plan with timeline estimates by problem type. Deduplication is typically the fastest operation; attribute standardisation and taxonomy reclassification take longer because they require agreed rules before the automated processing can run. Projects with multiple problem types are sequenced — deduplication first, taxonomy second, attribute standardisation third — so each stage builds on a cleaner baseline.
Platform choice doesn't change which catalog data problems need fixing — duplicates, broken taxonomy and missing attributes exist on Shopify, Magento and WooCommerce equally. The cleanup process is adapted to each platform's data model and export/import format. For Shopify: CSV export, deduplication in staging, reimport with merge logic. For Magento: database-level deduplication with attribute value normalisation via data feeds. For WooCommerce: CSV-based cleanup with plugin-level category reassignment. Amazon and eBay listings are cleaned against their respective listing format requirements. MercuryMinds has executed catalog cleaning on all major platforms since 2008.
Related Catalogue Services
Ready to Fix Your Catalog?
Tell us how many SKUs you have, which platform they're on, how you think the data debt accumulated and what's breaking most visibly right now. We start with a free data audit before scoping the cleanup.