Data governance
Data governance is the set of rules, roles, and approvals that decide who may create, change, or approve a product data value, and what a valid value looks like. For product data it answers four questions per attribute: who owns it, which source wins, what a valid value looks like, and what requires review. Without it, enriched catalogs drift back toward blanks and conflicts as new SKUs and supplier feeds arrive ungoverned.
What data governance decides
Governance is a set of decisions that show up as configuration in your PIM, ERP, and review queues.
For product data, governance answers four questions for every attribute:
- Who owns it. Which named role can set
thread_pitchon a fastener line, and which role can only suggest a value. - Which source wins. When the manufacturer's spec sheet says 600V and the ERP item master says 300V, what breaks the tie.
- What a valid value looks like.
Grade 8from a controlled list, not "grade eight" or "G8". - What requires review. A GTIN change on a live SKU gets approval. A marketing bullet rewrite does not.
Most catalogs already have implicit answers to these. They just live in one merchandiser's head, and leave when they do. Governance is the act of writing them down and enforcing them in the system rather than in the hallway.
A useful test: pick your worst attribute, then ask three people who owns it. If you get three answers, you don't have governance on that field.
A governance model for one attribute
Governance gets abstract fast. It stays concrete when you write it one attribute at a time. Here is what a filled-in row looks like for a UL listed 600V wire connector:
| Governance element | Example for voltage_rating |
|---|---|
| Owner | Product manager, electrical category |
| Trusted source | Manufacturer spec sheet PDF, current revision |
| Fallback source | UL certification directory listing |
| Valid format | Controlled list: 300V, 600V, 1000V |
| Who can propose | Enrichment vendor, supplier portal, category analyst |
| Who can approve | Product manager only |
| Review trigger | Any change on a SKU with live orders |
| Downstream blast radius | PDP spec table, faceted search filter, code-compliance claims |
Do this for 10 to 20 attributes that drive search, compliance, and returns. Governance applied evenly across every field is governance nobody follows.
The last row is the one that earns budget. voltage_rating is governed tightly because a wrong value is a safety claim. country_of_origin is governed tightly because it touches duty. marketing_bullet_3 can stay loose.
Governance vs. the terms it gets confused with
Buyers routinely ask for governance and get sold something adjacent.
| Term | What it is | Relationship to governance |
|---|---|---|
| Data governance | Rules, roles, approvals | The policy layer |
| Master data management | Reconciling records into one trusted version | The mechanism governance directs |
| PIM | Storage, modeling, workflow for product content | Where governance rules get enforced |
| Data quality | Measured state of the values | The scoreboard governance moves |
| Stewardship | The people doing the daily work | Governance staffed |
The common failure: a company buys MDM or a PIM, treats the purchase as the governance decision, and later discovers that the tool faithfully enforces rules nobody ever wrote. The system will happily let four teams overwrite each other if you never told it who wins.
Governance comes first logically, even when the software comes first on the purchase order.
Why enrichment decays without it
This is the part distributors feel most directly. A team runs a cleanup project, lifts fill rate on 40,000 SKUs, and the catalog drifts back toward where it started.
The decay is mechanical:
- New items arrive ungoverned. The cleanup covered existing SKUs. Nothing changed about how the next 3,000 get onboarded.
- Supplier feeds overwrite good values. A monthly feed pushes a blank or a worse value over a researched one, because no rule says the researched value wins.
- Nobody owns re-verification. A manufacturer revises the spec sheet for a 3/8-16 Grade 8 hex bolt. Your value silently goes stale.
- Exceptions become the norm. One rush deadline gets a bypass. The bypass becomes the path.
Governance stops the bleed by making each of these a rule with an owner: onboarding gates for new items, source precedence for feeds, a re-verification cadence for spec-sourced attributes, and an exception process that logs who approved the bypass.
The PIM stores your product data and holds the rules; Anglera does the work of completing the values so stewards approve rather than research.
Frequently asked questions
What is data governance for product data?
It is the rules and roles that control product attribute values: who owns each attribute, which source is trusted, what format is valid, and which changes need approval before they go live. Applied to a catalog, it is the difference between a spec value someone researched and a value someone typed. It is policy, enforced in your PIM and item-onboarding process.
Is data governance the same as master data management?
No. Governance is the policy layer: who decides, what rules apply, who approves. MDM is a mechanism for reconciling records from multiple systems into one trusted version. MDM executes what governance decides. Buying MDM without governance gives you a system that enforces rules nobody wrote, and four teams still overwriting each other inside a tool that was told nothing about who wins.
Who should own product data governance?
Ownership works best at the attribute level, assigned to whoever bears the consequence of a wrong value. Compliance attributes like voltage rating or country of origin belong to product or compliance managers. Merchandising copy belongs to marketing. Identifiers like GTIN and MPN usually belong to item setup. A single central owner for all fields tends to become a bottleneck everyone routes around.
How do you start governing a catalog that has none?
Start with 10 to 20 attributes that drive search, compliance, and returns. For each, write down the owner, the trusted source, the valid format, and what triggers review. Configure those rules in the PIM, then gate new item setup on them. Resist the urge to govern all 340 attributes at once. Broad governance nobody follows is worse than narrow governance that holds.
How do you know your governance is actually holding?
Look at the items onboarded since the rules went live rather than the ones you cleaned up. Measure fill rate and error rate on those SKUs against your governed baseline. Track how often the exception path gets used, how long approvals sit in the queue, and whether spec-sourced attributes get re-verified on the cadence you set. Governance that holds shows up as new items arriving already correct, with few bypasses to explain.