EPU Procurement Sustainability Intelligence

1. Purpose

This specification defines how EPU-linked product intelligence may support procurement decisions without collapsing commercial price, sustainability performance, modeled financial impact or accounting values into one misleading number.

The capability is decision intelligence. It does not change supplier invoice prices unless an independently governed commercial process explicitly does so.

2. Core principle

Comparability before ranking. Separation before aggregation.

ZAYAZ must establish that product sustainability results are sufficiently comparable before using them to rank procurement alternatives.

The EPU identity does not store price, PCF or adjusted cost. These are contextual commercial facts and governed analytical results connected to the product graph.

3. Procurement context

A procurement comparison should preserve references to at least:

buyer ECO
supplier ECO
EPU / product candidate
commercial offer
procurement unit
functional unit
price and currency
applicable PCF/result artifact
comparability decision
internal carbon-price policy
scenario/effective time
provenance

4. Comparability before ranking

ZAYAZ must not compare or rank footprint numbers merely because both are expressed as CO2e.

The Product Sustainability Comparability Resolver should evaluate at least:

functional unit
product specification / equivalence
system boundary
lifecycle stages
methodology / standard
allocation method
reference period
production geography
production context
primary vs secondary data
dataset / factor versions
biogenic / land-use treatment where relevant
uncertainty
verification state
TrustGate state

Canonical outcomes are defined by config/system/epu.procurement-intelligence.json:

COMPARABLE
CONDITIONALLY_COMPARABLE
NORMALIZABLE
NOT_COMPARABLE

Interpretation:

• COMPARABLE — governed comparison/ranking is permitted under the applicable policy.
• CONDITIONALLY_COMPARABLE — comparison/ranking is permitted only with the recorded limitations/conditions visible to the decision process.
• NORMALIZABLE — direct ranking is prohibited until a governed, methodologically permitted normalization has produced a new comparison basis.
• NOT_COMPARABLE — PCF-based ranking is prohibited for the supplied artifacts/context.

Missing or insufficient information is not a fifth comparability state. It is represented in the dimension-level evidence/outcome and results in NOT_COMPARABLE whenever the missing information prevents a defensible comparison. This keeps the machine contract finite while preserving the reason for failure.

A normalization must be methodologically permitted, versioned and reproducible. ZAYAZ must not create false comparability by silently transforming incompatible studies.

5. ComparabilityAssessment

Representative governed artifact:

ComparabilityAssessment
├── assessment_id
├── candidate_result_refs[]
├── product_scope_refs[]
├── functional_unit_profile
├── boundary_profile
├── method_profile
├── normalization_actions[]
├── comparability_outcome
├── limitations[]
├── uncertainty_notes[]
├── policy_version
├── generated_at
└── provenance_ref

Procurement ranking should reference the assessment used.

6. Carbon-Adjusted Acquisition Cost

For a simple product-level comparison:

CAAC = P + ((PCF_kgCO2e / 1000) × ICP_per_tCO2e)

Where:

• P = commercial purchase price per procurement unit;
• PCF_kgCO2e = comparable product footprint per aligned procurement/functional unit;
• ICP_per_tCO2e = organization's governed internal carbon price or shadow price.

Example:

Candidate	Price	PCF	ICP	Internal carbon cost	CAAC
A	€100	40 kgCO2e	€150/t	€6.00	€106.00
B	€103	12 kgCO2e	€150/t	€1.80	€104.80
C	€101	25 kgCO2e	€150/t	€3.75	€104.75

The lowest invoice price is not necessarily the lowest carbon-adjusted acquisition cost.

7. Financial semantic separation

ZAYAZ must preserve explicit distinctions between:

commercial_price
contractual_logistics_cost
modeled_logistics_cost
internal_carbon_cost
logistics_carbon_cost
modeled_risk_adjustment
modeled_compliance_adjustment
modeled_lifecycle_adjustment
carbon_adjusted_acquisition_cost
sustainability_adjusted_acquisition_cost

COMMERCIAL_PRICE and CONTRACTUAL_LOGISTICS_COST are actual/contractual value classes and may enter invoice/accounting flows where supported by authoritative source records.

MODELED_LOGISTICS_COST is scenario intelligence. It must never be represented as an invoice amount merely because a logistics scenario was calculated.

Internal shadow costs are decision variables, not supplier invoice amounts and not automatically accounting liabilities.

The canonical cost-component registry therefore prohibits an ACTUAL_OR_MODELED value class. A value must be semantically classified before it enters procurement intelligence.

Every adjusted result must identify which terms are:

ACTUAL_CONTRACTUAL
INTERNAL_SHADOW
MODELED_SCENARIO

Additional observation/estimate semantics belong in the referenced source/result provenance rather than weakening the financial value class.

8. Internal Carbon Price Policy

Internal carbon pricing should be a governed organization-level financial-impact policy rather than a free-form number entered into every comparison.

Representative context:

InternalCarbonPricePolicy
├── policy_id
├── organization_eco_id
├── currency
├── price_per_tCO2e
├── effective_from
├── effective_to
├── applicable_scopes
├── geography / business-unit overrides?
├── policy_basis
├── approved_by
└── provenance_ref

The policy may be scenario-specific and should not be represented as a statutory carbon price unless that is actually the policy basis.

9. Extended sustainability-adjusted cost

A broader scenario may use a model such as:

SAAC = PurchasePrice
     + ContractualLogisticsCost
     + InternalCarbonCost
     + ModeledLogisticsCost
     + LogisticsCarbonCost
     + RiskAdjustment
     + ComplianceAdjustment
     + LifecycleAdjustment

This is a decision model, not an accounting identity. Every component must retain its value class, method, uncertainty, source and policy lineage.

The UI must show the components separately before presenting an aggregate scenario value.

10. Uncertainty

PCF and modeled cost values may carry uncertainty. Procurement intelligence must preserve that uncertainty rather than rank candidates on false precision.

MICE may support deterministic, Monte Carlo, Bayesian or other governed scenario methods. Result lineage must identify the calculation run, engine/method version, inputs and applicable uncertainty model.

11. Trust and evidence

Procurement intelligence should incorporate TrustGate/evidence state. A lower PCF based on weak secondary assumptions must not be presented as equivalent in confidence to an independently verified supplier-specific result.

Trust does not replace comparability. A highly trusted result may still be methodologically incomparable with another result.

12. Non-goals

This capability does not by itself:

• alter supplier invoice prices;
• create accounting liabilities;
• execute purchases;
• create marketplace transactions;
• replace procurement approval policy;
• guarantee that two sustainability studies are comparable merely because normalization is technically possible.