Product data quality is measured across six dimensions: completeness, validity, accuracy, consistency, uniqueness and timeliness. Each needs a formula, a target and an owner. A PIM calculates completeness per family, channel and locale. A weighted score that combines several dimensions shows which products are ready to publish and where to direct enrichment effort.
Product data quality is managed through measurement. Teams track six dimensions (completeness, validity, accuracy, consistency, uniqueness and timeliness), and each needs a formula, a target and an owner. A PIM (product information management) system calculates completeness per family, channel and locale out of the box. A weighted score that combines several dimensions then shows which products are ready to publish and where enrichment effort will pay back first.
What are the dimensions of product data quality?
The UK Government Data Quality Hub, drawing on DAMA UK, names six dimensions. Each one maps to a concrete product data problem:
- Completeness: required attributes have values, such as a missing net weight blocking a retailer listing.
- Validity: values conform to type, format and range, such as a GTIN with a wrong check digit.
- Accuracy: values reflect the physical product, such as declared dimensions matching the measured pack.
- Consistency: values agree across attributes and systems, such as the colour in the title matching the colour attribute.
- Uniqueness: each product exists once, with no duplicate SKUs or GTINs.
- Timeliness: data is available when the channel needs it, such as before a launch date.
How do you calculate data quality metrics?
Each dimension needs a repeatable formula. The table shows common definitions with illustrative targets. Calibrate the targets to your own channels.
| Metric | Formula | Illustrative target |
|---|---|---|
| Completeness (per channel and locale) | Required attributes filled ÷ required attributes × 100 | 100% before publish |
| Validity rate | Values passing validation rules ÷ values checked × 100 | 99% or higher |
| Accuracy rate | Attributes matching a physical or source audit ÷ attributes audited × 100 | 98% or higher on audited samples |
| Consistency rate | Records with no cross-system mismatches ÷ records compared × 100 | 98% or higher |
| Duplicate rate | Duplicate records ÷ total records × 100 | Below 0.5% |
| Time to ready | Date channel-complete minus date created, in days | Agreed per category |
| Enrichment backlog | Products below threshold for a priority channel | Downward trend |
Completeness is usually native to the PIM. In Akeneo, for example, completeness depends on the family, the channel and the locale, and it is recalculated after imports, rules and saves. Accuracy is the hardest to automate, because it requires comparison with the real product. The GS1 Data Quality Framework includes a Data Inspection Procedure for physically validating attributes.
How should a weighted quality score work?
A single score helps teams prioritise, provided it stays transparent. A common illustrative model:
- Set channel completeness as a gate. Products below 100 per cent on mandatory attributes score zero for that channel.
- Weight the remaining dimensions, for example validity 30 per cent, enrichment completeness 30 per cent, consistency 20 per cent and asset coverage 20 per cent.
- Score by family and channel, because a fashion marketplace and a B2B portal value different attributes.
- Publish the score next to its components, so stewards can see what to fix.
Scores work best when they trigger action: workflow tasks, dashboard alerts or blocked exports.
Where should quality rules be enforced?
Prevention costs less than correction. Rules belong at four points:
- Data model: typed attributes, units, select lists and required flags, as covered in product data modelling.
- Input: validation on supplier imports and onboarding portals.
- Workflow: approval steps that block publication until checks pass.
- Output: channel-specific checks before syndication, such as retailer or GDSN validation rules described in our guide to GS1, GTINs and GDSN.
AI enrichment can fill gaps at scale, but generated values need the same validation and a human review sample.
Which standards support product data quality?
Three references are widely used. The GS1 Data Quality Framework contains a data quality management system, self-assessment tools and the inspection procedure. ISO 8000 is a multi-part international data quality standard. ISO 8000-110:2021 covers the exchange of master data as characteristic data, and separate parts address provenance (ISO 8000-120), accuracy (ISO 8000-130) and completeness (ISO 8000-140). DAMA's body of knowledge frames quality within wider data management.
Metrics only hold when someone owns them. The roles that do so are set out in product data governance.
Key takeaways
- Measure six dimensions: completeness, validity, accuracy, consistency, uniqueness and timeliness.
- Give every metric a formula, a target, a reporting frequency and an owner.
- Use channel completeness as a publication gate, then weight the other dimensions into a transparent score.
- Enforce rules in the data model, at input, in workflow and at output, and use reports to confirm they work.
- The GS1 Data Quality Framework and ISO 8000 provide external reference points for audits.
Frequently asked questions
How is product completeness calculated in a PIM?
Completeness is usually the number of required attributes that have a value, divided by the total number of required attributes, for a given product, channel and locale. Required attributes are defined per family and per channel, so the same product can be 100 per cent complete for a web shop and 70 per cent complete for a marketplace with stricter requirements.
What is a good product data quality score?
Targets depend on the channel and the attribute. Mandatory attributes for a channel should be 100 per cent complete before publishing, because many retailers and marketplaces reject incomplete records. Organisations typically set lower targets for enrichment attributes and raise them over time. Report trends weekly, since a single snapshot hides regressions.
Which standards define data quality for product data?
The GS1 Data Quality Framework provides a data quality management system, self-assessment tools and a physical data inspection procedure for trade item data. ISO 8000 is a multi-part international standard for data quality, with Part 110 covering the exchange of master data. The DAMA data management community describes the common quality dimensions.
For guidance on structuring a PIM selection, see the independent resource PIMselection.com.