One truth per product — why data quality decides omnichannel commerce

The observation. A single product record must today exist — accurate and consistent — across a dozen surfaces : the brand site, the online boutique, marketplaces, comparison feeds, reseller catalogues, the ERP, logistics. Yet in most organisations that information actually lives scattered — a spreadsheet here, a supplier PDF there, a description written three different ways depending on the channel. The problem is not a lack of data ; it is the absence of a single source of truth. That is precisely what Product Information Management (PIM) sets out to solve : ensuring there is only one place where product data is created, governed and published.

The quiet cost of bad data

Poor-quality data does not show up on a balance sheet — which is exactly what makes it dangerous. It is paid in returns driven by a misleading record, in support calls for a missing dimension, in lost ranking for unfilled attributes, in merchant feeds rejected by the platforms. A body of data-governance research estimates that organisations lose a meaningful share of revenue to bad data ; the formula popularised by analysts — the 1-10-100 rule — captures the intuition : preventing an error costs one, correcting it costs ten, suffering it costs a hundred.

That cost concentrates on the attributes that decide a purchase : material, dimensions, compatibility, certifications, what is in the box. A record with the right price but the wrong material does not produce a price complaint — it produces a return, a disappointment, and dented trust. Product-data quality is therefore less a technical matter than a relational one.

The five dimensions of quality

The master-data-management (MDM) literature breaks quality down into measurable dimensions. Five recur consistently :

• Completeness — is every mandatory attribute for a category filled in ? A chair with no seat height is unsellable online.
• Consistency — does the same fact say the same thing everywhere ? “Steel” on one channel and “metal” on another breaks trust and filtering.
• Accuracy — does the data match the physical reality of the product ?
• Timeliness — does it reflect the supplier's latest revision rather than a stale version ?
• Conformity — does it respect the expected standards (GS1, GTIN, units, marketplace category taxonomies) ?

Standards and syndication

Syndication — publishing one record to several channels with different requirements — is only sustainable on shared standards. That is the role of GS1, the body that maintains the GTIN (the global barcode) and the GDSN, the data-synchronisation network that lets a manufacturer and a distributor describe the same product without re-keying. Without a stable identifier and a common attribute vocabulary, each new channel becomes a manual mapping project — and every manual mapping is a source of error.

Modern PIM therefore does more than store : it also handles Digital Asset Management (DAM) — the media, photos, drawings, spec sheets, versioned and linked to the right reference. A low-resolution image, badly cut out or attached to the wrong colourway, does as much damage as a wrong attribute. Data and media are two halves of the same record.

Governance before the tool

A PIM is a tool ; quality is a discipline. The programmes that fail are almost always those that bought software without first defining who owns each attribute, who has the right to write it, and under what rule. Governance answers questions that are simple but decisive : does the material come from the supplier or the product team ? Can price overwrite a manual entry ? Is the translation a field of its own or a copy ? Without clear answers, two sources end up contradicting each other, and the tool merely accelerates the confusion.

The practical principle is field ownership : for each attribute, one and only one source is authoritative, the rest only read. It is less a technology question than an organisational one — and it is what separates a catalogue you trust from a catalogue you endure.

Where we stand

Montandor Andorra consolidates its product data through a PIM in that spirit : a single source for factual attributes, media versioned and tied to the right reference, and a clear rule on who owns which field — commercial data on one side, localised copy on the other. The aim is not completeness for its own sake ; it is that a customer, on boutique.montandor.fr or at a reseller, always reads the same accurate information.

“A product record is a promise. If it says ‘steel’ and we ship painted metal, we have lied — even unintentionally. Data quality is not an IT topic ; it is the most everyday form of our honesty toward the customer.”
— Wouter Meijboom, CEO, Montandor Andorra.

Sources

• GS1 — GTIN standards and the GDSN (Global Data Synchronisation Network), official GS1 specifications.
• DAMA International — DMBOK (Data Management Body of Knowledge), chapters on data quality and master data management (MDM).
• Foundational data-quality work : Wang & Strong, Beyond Accuracy: A Framework for Data Quality (Journal of Management Information Systems, 1996).
• Larry English — Improving Data Warehouse and Business Information Quality (1-10-100 cost rule).
• Ventana Research / ISG — studies on the cost of poor data quality in commerce.
• Gartner / Forrester — market analyses of Product Information Management and Master Data Management.

Research led by Chloé Garnier (Head of Architecture), in collaboration with Iga Zielińska (Visual Producer) and Mihail Ivanov (Integration Engineer).