Governance and lineage across the OEM-supplier-dealer chain

The data that actually flows

Follow the part number across three parties and you touch every major automotive dataset:

• Tier-1 supplier: engineering specs, the supplier part number, PPAP (Production Part Approval Process, the documented proof a part meets spec) records, and ASN (Advance Shipping Notice) data.
• OEM: the bill of materials (BOM, the structured list of every component in a vehicle), the internal part number, engineering change orders, and VIN (Vehicle Identification Number, the 17-character unique ID) build records.
• Dealer: the parts catalog, warranty claims, and repair orders (RO).

Each party assigns its own identifier. Nobody is wrong. They just never agreed on a shared key. That is the governance vacuum where dirty data breeds.

Data ownership: who decides what is true

Ownership is not "who stores the data." It is "who is accountable for a field being correct."

A useful rule: the party closest to the source of truth owns the field.

• The Tier-1 supplier owns the engineering dimensions and material spec. They made the part.
• The OEM owns the mapping between supplier part number and OEM part number, plus which vehicles the part fits (the fitment).
• The dealer owns the actual installation event: which VIN got which part on which date.

Write this down in a RACI (Responsible, Accountable, Consulted, Informed) matrix per field. When the caliper's torque spec changes, everyone knows the Tier-1 engineer is Accountable and the OEM parts team is Consulted. No committee, no email chain.

Master data managementMaster data managementMaster Data Management (MDM) is the discipline of creating and maintaining a single, consistent, trusted version of an organization's core business entities like customers, products, and suppliers.Voir la définition complète → and the golden record

MDMMDMMaster Data Management (MDM) is the discipline of creating and maintaining a single, consistent, trusted version of an organization's core business entities like customers, products, and suppliers.Voir la définition complète → is the discipline of creating one trusted version of a shared entity. That trusted version is the golden record.

For our caliper, the golden record answers: what is the single canonical part, and what are all its known aliases?

The technical move is entity resolution: matching BRBRThe percentage of visitors who leave after viewing only one page, often a signal of poor relevance, mismatched intent, or weak user experience.Voir la définition complète →-4471-A, 4471A, 04471-AA, and BR4471A-EU to one master ID. A minimal crosswalk table:

Now a recall query joins on master_part_id and catches all four aliases at once. Without it, the warranty system's 04471-AA is invisible to the recall built on 4471A.

Industry note: the GS1 standards provide global identifiers (like the GTIN) increasingly used to reduce exactly this alias chaos in automotive parts. Treat vendor-neutral standards as your default canonical key when one exists.

Lineage: proving where a value came from

Data lineageData lineageData lineage maps how data moves and transforms across systems, from origin to consumption, showing where it came from, what changed it, and where it goes.Voir la définition complète → is the documented path a value travels: source system, every transformation, and every destination.

Ask a concrete question: the golden record says the caliper's torque spec is 120 Nm. Where did that come from?

Good lineage answers instantly: Tier-1 CAD system → PPAP document v3 → OEM engineering import (2025-11-04) → part_master. Bad lineage means an engineer spends a day in emails and still guesses.

Lineage matters most during three events:

1. Recalls: regulators want the chain of custody for a defect. In the US that is the NHTSA (National Highway Traffic Safety Administration); in the EU, type-approval authorities under the framework regulation. They ask which VINs got the affected part. Lineage plus the alias table produces that list.

2. Warranty disputes: a Tier-1 supplier is charged for a failure. Lineage shows whether the failed part's spec matched what the supplier shipped, or whether the OEM changed the spec after approval.

3. Engineering changes: when the spec changes, lineage shows every downstream system that must be updated.

🎬 [VIDEO: "What is Data LineageData LineageData lineage maps how data moves and transforms across systems, from origin to consumption, showing where it came from, what changed it, and where it goes.Voir la définition complète →?" - youtube.com - a clear 6-minute primer on source-to-target tracing and why it matters for audits]

Stewardship KPIs that keep the record trustworthy

A golden record decays the moment you stop measuring it. A data stewarddata stewardA business-side owner responsible for the quality, consistency and appropriate use of data in their domain.Voir la définition complète → is the named person accountable for quality in a domain (parts, VINs, dealers). Give them KPIs, not vibes.

Core stewardship metrics, with illustrative targets. Treat all numbers below as example targets to negotiate internally, not industry benchmarks:

• Match rate: share of incoming part records auto-resolved to a master ID. Target above 95 percent. Below that, humans drown in manual matching.
• Duplicate rate: share of master records that are actually the same part. Target below 1 percent.
• Completeness: share of golden records with all mandatory fields (canonical name, fitment, owning party) populated. Target above 98 percent.
• Timeliness: median lag from a source change to golden-record update. Target under 24 hours for safety-relevant fields.
• Stewardship response time: median time to resolve a flagged data issue. Target under 3 business days.

A worked calculation: match rate

Suppose in one month your Tier-1 ingestion receives 40,000 part records. Entity resolution auto-matches 37,600 to an existing master ID. A steward manually matches 1,800 more. 600 remain unmatched (new or ambiguous parts).

Auto match rate = 37,600 / 40,000 = 94.0 percent.

That is just below a 95 percent target. The 2,400 non-auto records are your steward's workload. If each manual match takes 4 minutes, that is 1,800 × 4 = 7,200 minutes, roughly 120 steward-hours that month. Now you can staff the function with real numbers instead of guessing.

Track match rate over time. A sudden drop usually means a source system changed its format without telling you, which is a lineage and governance failure you want to catch in days, not quarters.

The balance of power (and why governance is hard)

Governance is a data problem wrapped in a power problem.

OEMs sit at the center and have historically dictated data formats to suppliers. Large automakers (Volkswagen, Toyota, General Motors, Stellantis, Ford, and others) run supplier portals where Tier-1s must submit data in the OEM's schemaschemaA schema is the formal blueprint that defines how data is structured, named, typed, and related within a database, file, or message.Voir la définition complète →. That gives the OEM leverage to define the canonical part number.

But dealers are often independent businesses, not owned by the OEM. In the US, franchise laws (state-level dealer franchise statutes) protect that independence. So the OEM cannot simply command dealer data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.Voir la définition complète →. It has to incentivize it, often by tying warranty reimbursement to clean repair-order data.

The practical result: the OEM usually owns the master part identity, the Tier-1 owns the engineering truth, and the dealer's cooperation must be earned through data-sharing agreements. Your governance model has to respect that no single party controls the whole chain.

Regulation that forces the discipline

Two real regimes raise the stakes:

• Vehicle safety recalls: NHTSA (US) and EU type-approval authorities can compel traceability from part to VIN. Weak lineage means a slow, over-broad recall, which is expensive and reputationally damaging.
• Data protection: repair orders contain personal data (owner name, contact). In the EU, the GDPR (General Data Protection Regulation) applies. Your lineage must show where personal data lives so you can honor deletion requests without breaking the parts record.

Governance is not paperwork. It is what lets you answer a regulator in hours.

Putting it together

For one caliper, the full loop looks like this: the Tier-1 owns and submits the spec, MDMMDMMaster Data Management (MDM) is the discipline of creating and maintaining a single, consistent, trusted version of an organization's core business entities like customers, products, and suppliers.Voir la définition complète → resolves the four aliases into master ID MP-100293, lineage records the source of every field, and a steward watches the KPIs so the golden record stays above target. When the recall comes, one query on the master ID reaches every alias, every VIN, and every dealer.

That is the difference between a controlled recall and a chaotic one.

Key Takeaways

• One physical part, many names. The core job is entity resolution: mapping every supplier, OEM, and dealer alias to a single master part ID so recalls and warranty queries never miss a record.
• Ownership means accountability per field, not storage. Assign each field (spec, fitment, installation event) to the party closest to its source, and record it in a RACI matrix.
• Lineage is your audit answer. Documented source-to-target paths let you satisfy NHTSA, EU authorities, and GDPR requests in hours, and settle warranty disputes with evidence.
• KPIs keep the golden record alive. Track match rate (target above 95 percent), duplicate rate, completeness, and timeliness. All targets here are illustrative examples to set internally, not published benchmarks.
• Governance respects the power balance. OEMs own master identity, Tier-1s own engineering truth, and independent dealers must be incentivized, not commanded, into clean data.