What Is a Golden Record in PIM.RED? A Real Example From Stämpfli
The problem: every source owns a different piece of the truth
Stämpfli, a Swiss publisher of legal and professional content, manages a product catalog that pulls data from six distinct sources.
Two are internal:
Stämpfli’s own backend system, which exports product data via SMPT
PIM.RED itself, where editorial teams can author and define products directly, making PIM.RED both a destination for data and a source of it
Four are external, tied to the broader Swiss and German book trade infrastructure:
Ceebo, which provides eBook-specific metadata
Buchzentrum (BZ), the central Swiss book trade data platform
AVA, a Swiss book distribution and logistics provider, configured early in the integration but not actively used today
KNV, the legacy German book trade data feed, now routed through Zeitfracht, also inactive but still mapped in the system
Here’s the core issue: none of these sources has the full picture. Each one is authoritative for a narrow slice of the product and largely blind to the rest.
BZ can confirm that a title exists and is available through Swiss trade channels, but its descriptive metadata is often thin. Ceebo holds detailed eBook attributes, format, file type, DRM information, but knows nothing about print editions. Stämpfli’s internal system holds the editorial source of truth, the actual content and classification, but doesn’t carry distribution or trade metadata at all.
So a single product might technically "arrive" through BZ, while the richest data about that same product actually lives in Ceebo. Neither source is wrong. They’re just incomplete in different ways.
Why you can’t merge everything
The instinct, when facing multiple data sources, is to combine them. Pull in every field from every system and stich together one giant record.
This doesn’t work, for a simple reason: when two sources disagree, something has to decide which one wins.
If BZ says a book is in print and Ceebo says it’s eBook-only, that’s not a data quality problem you can solve by importing both values. It’s a conflict that needs a rule. If Stämpfli’s internal system updates a title’s classification but the external feed hasn't caught up yet, PIM.RED needs to know which value is current.
Merging everything without resolution logic doesn’t create a single source of truth. It creates a record with multiple, sometimes contradictory truths side by side. That’s arguably worse than the spreadsheet chaos it was supposed to replace, because now the inconsistency is hidden inside a system everyone assumes is reliable.
The golden record: the resolution, not the collection
A golden record isn’t a copy of everything. It’s the result of a deliberate decisions: for each attribute, which source is authoritative and how conflicts get resolved.
In practice, this means defining a field-level hierarchy. For a given product, the editorial description might always come from Stämpfli’s internal system. The eBook format details always come from Ceebo. Trade availability always comes from BZ. When a new product enters the pipeline through BZ but its richest data originates from Ceebo, the system pulls in exactly the fields it needs from Ceebo, according to the rules already defined, and leaves the rest alone.
The result is a single, de-duplicated product file that combines the most accurate, complete data from the best sources for each detail.
Why this matters more than the catalog itself
The catalog, storefront, and search are the visible outcomes of a PIM.RED, but the real core is the conflict resolution logic and determines which source wins on each attribute when systems clash.
Get that logic right, and everything built on top of it, searching indexing, channel exports, metadata feeds, simply works, because it’s operating on data that’s already been resolved.
Get it wrong, and you’ve built a more expensive version of the same inconsistency that spreadsheets created in the first place. The interface looks more sophisticated. The underlying problem is identical.
Not every PIM handles this at the field level. Many let you define a source priority per feed, but resolving conflicts attribute by attribute, with rules that persist across every sync, is a different capability. It’s also the part of PIM.RED’s architecture that receives the least attention and does the most work.
The golden record is what separates the two outcomes.