Product teams may well treat product data with a certain mistrust because it’s “messy,” or “unreliable.” But if they really dig down, they’ll realise they mistrust it because their business has neither an ownership model or a validation loop. It’s that unfortunate combination which guarantees unnoticed errors, conflicting versions of ‘the truth,’ and repeated embarrassment when dealing with irritated customers and stakeholders.
Our article outlines the specific failure pattern behind this lack of confidence in data quality, the operational ramifications it causes, and finishes with the best corrective sequence to re-establish that much-needed trust: STABILISE, STANDARDISE, ENFORCE.
The triple failure: No owner No gate. No proof
In most mid-to-large digital merchants (be they online uniquely, or hybrid), that foundational element, product data, tends to be touched by everyone but owned by nobody. Each siloed team beavers away at its own speciality – nothing wrong with that, sure. But it creates a risky scenario:
- Merchandising creates records
- Content enriches copy
- Supply chain adjusts weights and dimensions
- Purchasing updates prices
- Suppliers overwrite fields within their feeds.
- IT ostensibly “runs the system” but doesn’t actually own correctness.
Accountability is so opaque, so diffuse. Where no-one is ultimately responsible for the product record being correct from end-to-end, the outcomes are predictable:
- Values exist with no visible provenance (as in where it came from, when it was altered, who actually approved it)
- Validation is both optional and inconsistent (leading to the system allowing users to save incomplete or contradictory records)
- Quality approval gates are unclear or bypassed (as in the “we’ll fix it after go-live” syndrome)
Once teams cannot explain how a wrong value got into the record, the whole system becomes a kind of free-for-all. Moreover, it only takes one high-profile failure (maybe the wrong dimensions, lack of compatibility, or incorrect compliance document content), and any trust in the usability of the data collapses – a dysfunctional set of circumstances that won’t solve itself on its own.
The operational consequence: shadow processes become the real process
Low confidence tends to foster the same behaviour: Given the lack of a viable alternative, people build workarounds.
- “Final_v7” spreadsheets that become the ‘true’ master.
- Endless manual cross-checking against supplier PDFs or websites.
- Private image folders on various desktops because the supposedly definitive DAM/PIM asset set is unreliable.
- Flurries of Slack messages back and forth like “Can you do me a favour and check this SKU before I publish?”
It isn’t a question of stubbornness. It is improvised risk management by people who don’t have an alternative solution. However, it has damaging impacts on efficiency, productivity, and commercial credibility. At a minimum:
- Duplicated effort
- Long lead times
- A permanent backlog of “data debt” which blocks timely launches
The commercial and risk impact: speed, revenue, returns, compliance
When teams don’t have confidence in the product data they’re handling, the business pays in three ways:
- Time-to-market delays: products sit in ‘draft-form limbo’ because no one dares to sign off a record they can’t defend.
- Revenue leakage: channel listings get rejected, search relevance drops, and the conversion rate suffers when attributes are incomplete or inconsistent.
- Returns and compliance exposure: inaccurate dimensions, incomplete materials composition, or lack of reliable compatibility, or safety information. They all drive up avoidable returns, customer complaints, and risk of regulatory non-compliance.
“More training!” or installing a new tool is not the solution if there’s no ownership or validation loop for product data.
Why this persists even if you have a PIM
A PIM can store data, but it doesn’t automatically create trustworthy data. That foundational trust in usable data needs three elements which many merchants still fail to implement:
- Ownership at attribute level: “the PIM team” is too vague – you need in place named and domain-specific data stewards (like Technical Specs, Compliance, Logistics, and Marketing Content)
- Enforced validation rules: use mandatory attributes by, for instance:
ü Channel and category
ü Unit-of-measure controls
ü Permitted values
ü Rules on dependencies (such as: if “Battery Included = Yes,” then “Battery Type” spec is mandatory)
- Visible lineage and an audit trail: every value needs its source, timestamp, editor, approval status, and reason for any change(s) made.
Without these, a PIM essentially becomes a glorified filing cabinet. Teams still have to manually verify everything, destroying ROI and keeping confidence low in the process.
The missing feedback loop: outcomes never make it back to the record
The most damaging information gap is an absence of feedback. It’s the silo problem: A product return caused by wrong dimensions gets logged in customer service. A marketplace suspension is handled by the operations team. A flood of “does this fit?” queries are dealt with by support. But the product record often remains unchanged because there’s no reference point to see what’s happening.
If failures in data quality aren’t traced back to the exact SKU and attribute, nothing will improve. The consequences?
- Suppliers aren’t corrected
- Internal users aren’t coached
- Validation rules aren’t tightened
- Errors are repeated, and mistrust becomes the user mindset
Fix it in the only sequence that works: STABILISE, STANDARDISE, ENFORCE
1) Stabilise: ‘stop the bleeding’
- Define your master record: That is, which system is authoritative for which attributes (like ERP for cost and stock, PIM for enriched attributes, WMS for pack dimensions, and so on)
- Implement minimum viable approval gates for publish-critical fields.
- Activate clearly visible audit checking indicators: This involves showing “source,” “last updated,” “changed by,” and “approved by” on the product record
- If supplier feeds have overwritten trusted values without review, quarantine this data for remedial measures.
2) Standardise: make records comparable
- Build an attribute dictionary: Include at minimum, definitions, formats, units, allowed values, and examples
- Create supplier templates which are aligned to your schema, not theirs (in other words, category-specific and with mandatory fields and valid-value lists)
- Rationalise category structures and remove unused fields that create noise and mistakes
- Establish an enrichment workflow with clear protocols for hand-off (such as draft → enrich → validate → approve → publish)
3) Enforce: make quality the default
- Implement validation rules which will block save or publish if and when stipulated conditions aren’t met.
- Add data quality scoring at SKU level (as in completeness, conformity, consistency, freshness) and make it visible to teams.
- Link data quality to incentives: include measurable quality KPIs in role scorecards, which is where the work happens.
- Close the loop: trace returns, complaints, rejections, and suspensions back to specific attributes so you can trigger corrective measures.
Next step: get to the root cause, fast
If your teams are still double-checking SKUs across spreadsheets, it’s practically certain it’s because of ownership and validation flaws, not a people problem. Get in touch with us today at Start with Data for a discovery call. We can then arrange to map your master record, ownership model, and validation loop so your product teams don’t spend all day chasing their tails and put their talents to more strategic use and publish product information with confidence.