Definition

Data lineage records where data originated, how it moved and changed, and where it is used. It connects a reported result to the source records, transformations, mappings, and calculations that produced it.

For procurement and supply-chain analysis, lineage helps answer a practical question: can the team explain this number well enough to use it in a material decision?

Trace more than the file name

Knowing that a dashboard came from an ERP extract is only a starting point. The result may also depend on currency conversions, supplier mappings, category rules, exclusions, and manual adjustments.

VPP recommends retaining the transformations that materially affect interpretation. A reviewer should be able to reconstruct why the reported value differs from the original transaction.

  • Source: System, record identity, extract timing, and scope.
  • Transformation: Cleaning, conversion, normalization, and calculation rules.
  • Mapping: Supplier identities, category assignments, and organizational relationships.
  • Adjustment: Manual changes, approvals, and explanations.
  • Version: The rules and data snapshot used for a particular result.
  • Destination: Reports, models, and decisions that depend on the output.

Lineage should be proportionate to the decision. A material savings baseline requires more than an undocumented spreadsheet total.

A hypothetical category-total disagreement

Assume Finance reports $12 million in a category while a sourcing dashboard reports $10.8 million. Without lineage, the teams may debate which system is correct.

The trace shows that the sourcing view excluded $700,000 of taxes and $500,000 of transactions outside the event’s addressable scope. The $1.2 million difference is explained, but that does not automatically make either view appropriate for every purpose.

The teams should confirm which scope each decision requires and label the measures accordingly. A transparent difference can be governed; an unexplained difference undermines confidence.

If the exclusions are later changed, preserve the earlier version used to approve the event. Recomputing history without a record of the change makes later benefit validation difficult.

Use lineage to manage change

A revised supplier hierarchy or classification rule can affect several dashboards and sourcing models. Lineage helps identify which outputs depend on the changed element.

Before applying a broad update, assess its impact and decide which historical results should remain fixed. The appropriate treatment depends on whether the change corrects an error, reflects new facts, or introduces a new reporting policy.

Record the decision and communicate it to users. A technically correct pipeline can still mislead if a metric changes meaning without notice.

Keep lineage distinct from data quality

Lineage explains how a result was produced. It does not guarantee that the source data or transformation logic was correct.

Pair lineage with validation, reconciliations, and ownership. A fully traceable error remains an error, but it is easier to locate and correct.

Similarly, access to the trace should respect sensitive information. Users may need to understand a calculation without receiving unrestricted access to every underlying record.

Apply the same discipline to AI-assisted analysis

Record where model-generated classifications or matches enter the workflow and how they are reviewed. Preserve relevant versions, rules, and overrides when they affect material conclusions.

Do not describe an opaque generated explanation as lineage. The business needs an actual connection to records and processing steps, not a plausible narrative about how a result might have been created.

Frequently asked questions

  • Is lineage the same as provenance? They are related; provenance emphasizes origin, while lineage commonly follows the subsequent flow and transformations.
  • Does lineage prove a number is accurate? No. It supports verification but must be paired with quality controls.
  • Does every user need every source record? No. Provide appropriate traceability while enforcing access restrictions.

Make the analysis reviewable

VPP connects spend intelligence with the evidence needed for commercial and financial decisions. A useful result should be explainable through its records and transformations, not just persuasive in a dashboard.