AI System of Record and the 42-view contract

Venturi is the system of record for AI: an authoritative, replayable record of AI activity, accountable ownership, connected workflows and systems, normalized cost, and accepted business outcomes. Customers choose domain offerings around the decisions they need to make. Workflow Economics Core supports economic decisions without being a universal first purchase or prerequisite for other domains.

Customer promise: One place to understand how your company uses AI.

Each measurement preserves its source, scope, denominator, freshness and evidence state. Missing inputs remain visible, and numerical customer results require supporting records.

Source and epistemic boundary

The canonical view manifest is 17-AI-SYSTEM-OF-RECORD-VIEW-MANIFEST.json, copied from the Corpus positioning source. Its SHA-256 is 507cce71385b2300531d163747d0eda4443c181476c666ff9f5ba31c983ff6c9. The owner supplied the following briefs; their contents define the positioning and view design; numerical customer results retain their own evidence:

Source	SHA-256	Boundary
AI System of Record Brief.docx	3cf3080c0eac7d1afd99de6fe6a69b153e94d382c258165d7fa69a04712fb0bb	owner-supplied positioning and commercial design; customer results have separate evidence
What would be the top 25-50 metrics_measures_view....docx	72bf479aa8ba749bf25ff9c7cda5a40e4ba647f6b4107d0892f829dfb260d52c	owner-supplied metric views mapped to existing catalog IDs; values depend on their declared evidence

The record and its six economic layers

The record joins six layers for each eligible event or run:

Layer	What it establishes	Feasible read-only evidence
Identity and accountability	Who or what is responsible	Identity provider, service account, team directory, ownership registry
Workflow and orchestration	Which workflow, agent, job, or run produced the activity	Workflow registry, orchestration events, scheduler and deployment telemetry
Connected system	Which application, repository, data system, or downstream system participated	Service catalog, repository metadata, traces, deployment records, system logs
Model, provider and seat	Which model/provider, account, seat, or entitlement was used	Provider usage and audit APIs, gateway telemetry, seat and entitlement records
Normalized cost	What the eligible attempt cost, including paid failure and retry work	Billing exports, usage records, price books, allocation metadata
Accepted business outcome	What the customer accepted as the result	Customer outcome receipt, workflow status, review record, transaction or ticket record

These measurement views use read-only, out-of-band connectors to existing systems of record. They do not require raw prompt or completion payloads. The optional drop-in proxy is a separate acquisition mode that receives request and response content transiently on the request path; attribution processing remains asynchronous. Customer traffic fails open. Inferred operational confidence is capped at 0.95; unknown remains unknown and never becomes zero. Paid failures, retries, rejected work, and rework remain in the cost boundary. Cash savings and released capacity are separate value types.

Evidence and denominator rules

• Every view declares its decision, formula, unit, eligible denominator, source classes, customer inputs, and availability rule.
• Ratios are unavailable when the eligible denominator is zero or unknown. A missing input is not a zero value.
• Evidence states are observed, derived, estimated, unavailable, and not_admitted.
• Customer acceptance rules define the business denominator. Venturi does not substitute a generic success proxy for customer acceptance.
• The existing 70 domain measures and 46 ROI measures remain the ontology. These 42 views are decision-oriented views mapped to those stable IDs.

View families

Family	Views	Decision	Primary stakeholders
Foundational Spine	8	Can the customer trust the attribution record?	FinOps, Engineering, Risk
Workflow Economics Core	10	Which workflows should be scaled, changed or stopped?	Finance, FinOps, Workflow owners
Adoption Intelligence	6	Where is accepted AI work becoming embedded across teams?	AI program leaders, Product, Engineering
Quality and Reliability	5	Does the workflow produce accepted work reliably at its observed cost?	Product, Engineering, Operations
Security and Governance	6	Can the organization prove safe, accountable and compliant use?	CISO, Legal, Risk, Audit
Capacity and Infrastructure	4	Can capacity, latency and commitments support the workload?	Infrastructure, Platform Engineering, FinOps
Sustainability	3	What energy and emissions are attributable to accepted AI work?	Sustainability, Infrastructure, Finance

All 42 source-bound views

The table below is generated from the canonical manifest. Existing metric IDs are references to the existing catalog, not new metric identities.

ID	View	Family	Existing IDs	Formula	Unit	Eligible denominator
ASOR-FND-01	Six-Layer Attribution Graph View	foundational_spine	METRIC-QUALITY-03; METRIC-ROI-38; METRIC-ROI-39	join(identity, workflow, connected_system, model_provider, cost, accepted_outcome) for each eligible event or run	topology and coverage	eligible observed events or runs
ASOR-FND-02	Estate Coverage Ratio	foundational_spine	METRIC-COST-03; METRIC-QUALITY-07; METRIC-ROI-38; METRIC-ADOPTION-01; METRIC-ROI-37	covered eligible AI spend / eligible AI spend, reported separately from covered eligible invocation volume / eligible invocation volume	ratio	eligible AI spend or invocations in the declared source window
ASOR-FND-03	Attribution Confidence Score	foundational_spine	METRIC-QUALITY-01; METRIC-ROI-39	operational confidence c_oper for each resolved ownership link, with c_oper <= 0.95	score	resolved ownership links
ASOR-FND-04	Unresolved Volume, Unknown Stays Unknown	foundational_spine	METRIC-COST-06; METRIC-ROI-14; METRIC-ROI-39	Report unresolved eligible spend in USD and unresolved event count as absolute observations; separately report unresolved spend / eligible spend and unresolved events / eligible events when each denominator is positive, known and compatible.	USD, count and ratio	eligible spend or events
ASOR-FND-05	Shared-Key and Machine-Account Disaggregation Rate	foundational_spine	METRIC-QUALITY-01; METRIC-ROI-39	shared-key or machine-account events resolved to an accountable workflow / eligible shared-key or machine-account events	ratio	eligible shared-key or machine-account events
ASOR-FND-06	Source Telemetry Freshness and Lag View	foundational_spine	METRIC-QUALITY-04; METRIC-ROI-39	METRIC-ROI-39.source_to_observation_lag = observed_at - source_event_at; METRIC-QUALITY-04.ingest_to_served_lag = served_at - observed_at; METRIC-QUALITY-04.event_to_served_lag = served_at - source_event_at, reported separately by source and matched revision	duration	received source records
ASOR-FND-07	Owner Correction and Override Log	foundational_spine	METRIC-QUALITY-10; METRIC-ROI-39; METRIC-ROI-43	append-only correction and override events ordered by effective time and transaction time	replayable event history	correction events in scope
ASOR-FND-08	Undeclared or Shadow AI Spend Discovery	foundational_spine	METRIC-SECURITY-01; METRIC-ROI-30; METRIC-ROI-39	discovered unmanaged AI spend grouped by channel, provider and accountable aggregate owner	USD	discovered AI spend
ASOR-WE-01	Cost per Accepted Result	workflow_economics	METRIC-ROI-01; METRIC-WORKFLOW-02	total supported workflow cost / mature accepted downstream outcomes	USD per accepted result	mature accepted outcomes under the customer acceptance rule
ASOR-WE-02	Gross Cost per Invocation vs Effective Cost per Accepted Result	workflow_economics	METRIC-COST-02; METRIC-ROI-01; METRIC-ROI-14; METRIC-ROI-23	gross supported invocation cost / eligible invocations compared with total supported workflow cost / mature accepted outcomes	USD per invocation and USD per accepted result	eligible invocations and mature accepted outcomes
ASOR-WE-03	Fully Loaded Workflow Cost Breakdown	workflow_economics	METRIC-ROI-01; METRIC-COST-01; METRIC-ROI-18	Sum disjoint model, embedding, retrieval, orchestration, tool, infrastructure, human review, human rework and allocated support cost components within the declared workflow cohort.	USD by component	declared workflow cost boundary
ASOR-WE-04	Net Realizable Benefit Bridge	workflow_economics	METRIC-ROI-02; METRIC-ROI-33	Non-overlapping attributable incremental financial benefit relative to a compatible counterfactual - complete incremental investment in the same period; released capacity valuation remains a separate scenario.	USD	not a ratio; benefit and cost periods must align
ASOR-WE-05	Cash Savings vs Released Capacity Split	workflow_economics	METRIC-ROI-03; METRIC-ROI-04	net_labor_equivalent_hours = baseline_equivalent_hours - all_current_human_hours - disjoint_setup_maintenance_and_other_overhead_hours; cash reduction, AI-attributable released hours and evidenced redeployed hours retain their separate governed output contracts	USD and hours	separate cash and labor cohorts
ASOR-WE-06	Avoidable Workflow Cost	workflow_economics	METRIC-ROI-14; METRIC-QUALITY-05	Report disjoint paid failed-attempt, retry, abandoned and duplicate-work costs, including retry-rescued attempts; classify an avoidable subset only under an evidenced prevention scenario that preserves accepted outcomes.	USD	eligible workflow cost
ASOR-WE-07	Budget Owner Chargeback and Showback Ledger	workflow_economics	METRIC-COST-07; METRIC-ROI-40; METRIC-ROI-37	direct cost + approved shared allocation by effective owner and cost center	USD by owner and period	reconciled covered cost
ASOR-WE-08	Recurring Reconstruction Effort Saved	workflow_economics	METRIC-ROI-40; METRIC-ROI-04	baseline close effort - current close effort, with loaded labor valuation reported separately	hours and USD scenario	comparable close periods and declared roles
ASOR-WE-09	Workflow Economics Decision Brief View	workflow_economics	METRIC-ROI-01; METRIC-ROI-02; METRIC-ROI-12; METRIC-ROI-33	render scoped cost, accepted volume, drivers, alternative, assumptions, evidence and recommendation from one versioned cohort	decision artifact	declared workflow cohort and decision period
ASOR-WE-10	Covered AI Spend Tier Tracker	workflow_economics	METRIC-COST-03; METRIC-COST-09; METRIC-CAPACITY-07; METRIC-ROI-37	covered eligible AI spend by contract tier and trailing period	USD and tier	eligible covered spend
ASOR-AD-01	Workflow-Level Adoption Funnel	adoption_intelligence	METRIC-ADOPTION-03; METRIC-ADOPTION-06; METRIC-ROI-26; METRIC-ROI-28	count of approved workflows by stage: noticed, trialed, repeated, embedded, accepted	count and conversion ratio	approved workflow cohorts
ASOR-AD-02	Licensed Seat vs Active Workflow Utilization	adoption_intelligence	METRIC-ADOPTION-05; METRIC-ADOPTION-07; METRIC-ROI-20	active outcome-producing seats or workflows / licensed seats or workflows	ratio	licensed seats or workflows in scope
ASOR-AD-03	Accepted-Work Adoption Rate	adoption_intelligence	METRIC-QUALITY-06; METRIC-ROI-26	accepted target tasks with AI assistance / eligible target tasks	ratio	eligible target tasks in the declared business cohort
ASOR-AD-04	Human-in-the-Loop Override and Abandonment Rate	adoption_intelligence	METRIC-ROI-09; METRIC-ROI-07; METRIC-ROI-26	count_unique(eligible_mature_terminal_output_units_with_qualifying_override_or_abandonment) / count_unique(eligible_mature_terminal_output_units)	ratio	distinct eligible AI output decision units first delivered in the reporting period, with observable terminal actions and mature versioned windows
ASOR-AD-05	Cross-Departmental Capability Diffusion Map	adoption_intelligence	METRIC-ADOPTION-03; METRIC-ADOPTION-04; METRIC-ROI-27; METRIC-ROI-01; METRIC-ROI-26	accepted workflow volume and unit economics by aggregate business unit, role group and geography	map, count and ratio	eligible accepted workflow events
ASOR-AD-06	Program Mandate vs Realized Production Pull	adoption_intelligence	METRIC-ROI-30; METRIC-ROI-32; METRIC-ROI-33	verified recurring accepted workflow volume / declared program target volume	ratio and trend	declared program target
ASOR-Q-01	End-to-End Task Acceptance Rate	quality_reliability	METRIC-QUALITY-06; METRIC-WORKFLOW-06	mature runs whose downstream record reaches the declared accepted state / eligible mature runs	ratio	eligible mature runs
ASOR-Q-02	Retry, Fallback and Loop Amplification Factor	quality_reliability	METRIC-ROI-14; METRIC-ROI-16; METRIC-QUALITY-09; METRIC-ROI-01	total model and tool calls in a run / minimum nominal calls for the declared workflow	factor	declared nominal model and tool call count for the same workflow run
ASOR-Q-03	Human Review and Correction Burden	quality_reliability	METRIC-ROI-04; METRIC-ROI-07; METRIC-ROI-29; METRIC-ROI-09	review and correction minutes / mature accepted outputs	minutes per accepted output	mature accepted outputs with supported review evidence
ASOR-Q-04	Model and Provider Swap Impact Matrix	quality_reliability	METRIC-ROI-13; METRIC-ROI-23; METRIC-QUALITY-06	paired route cost per accepted result and acceptance rate for comparable cohorts	matrix of USD/result and ratio	matched workflow cohorts by route
ASOR-Q-05	Drift and Outcome Degradation Alert View	quality_reliability	METRIC-QUALITY-06; METRIC-QUALITY-09; METRIC-ROI-25; METRIC-ROI-01	accepted-result rate and cost per accepted result by versioned cohort over time, with declared change points	trend and alert	mature comparable workflow cohorts
ASOR-SG-01	Decision Provenance Reconstruction View	security_governance	METRIC-SECURITY-07; METRIC-ROI-39; METRIC-ROI-43	lineage chain from actor and workflow through model, cost, policy and accepted decision	audit lineage	eligible decisions or scoped cost
ASOR-SG-02	Shape Advisory and Gate Policy Status	security_governance	METRIC-SECURITY-06; METRIC-ROI-43	policy evaluations, advisory findings, availability gaps and signed customer opt-in state	status and counts	eligible policy evaluations
ASOR-SG-03	Decision-Time Policy Violation Rate	security_governance	METRIC-SECURITY-06; METRIC-ROI-43	unique eligible decision-policy evaluations with at least one breach / eligible decision-policy evaluation count	ratio and count	eligible decision-policy evaluation count
ASOR-SG-04	Deployment Boundary and Zero-Egress Verification	security_governance	METRIC-SECURITY-03; METRIC-SECURITY-04; METRIC-ROI-43	conforming checked paths / eligible checked paths, plus unchecked path count and continuity evidence	ratio, count and status	eligible checked processing, egress and retention paths
ASOR-SG-05	Regulatory Readiness Map	security_governance	METRIC-SECURITY-07; METRIC-SECURITY-08; METRIC-ROI-42; METRIC-ROI-43	workflow and control evidence coverage by declared risk tier and control requirement	map and ratio	eligible workflows and control requirements
ASOR-SG-06	Third-Party Model and Subprocessor Exposure Inventory	security_governance	METRIC-SECURITY-01; METRIC-SECURITY-07; METRIC-ROI-39	distinct observed providers, endpoints, subprocessors and jurisdictions by workflow and period	inventory and count	observed AI channels in scope
ASOR-C-01	Multi-Provider Traffic and Quota Utilization	capacity_infrastructure	METRIC-CAPACITY-01; METRIC-CAPACITY-02; METRIC-CAPACITY-03; METRIC-ROI-36	observed requests or tokens divided by compatible provider quota for each provider, account, model, region and window	count, tokens and ratio	fresh compatible quota
ASOR-C-02	End-to-End Workflow Latency Decomposition	capacity_infrastructure	METRIC-CAPACITY-05; METRIC-CAPACITY-06; METRIC-ROI-24	end_to_end_duration = terminal_at - started_at; stage contributions partition non-overlapping intervals on the critical path, with parallel and nested span work reported separately	duration by component	eligible workflow runs with complete timing metadata
ASOR-C-03	Committed Cloud and Provisioned Throughput Burn-Down	capacity_infrastructure	METRIC-CAPACITY-07; METRIC-CAPACITY-10; METRIC-ROI-22	committed usage and spend coverage, realized commitment delta and idle committed-capacity cost by period	ratio and USD	eligible commitment scope
ASOR-C-04	Fail-Open and Degraded-Mode Telemetry Continuity	capacity_infrastructure	METRIC-CAPACITY-05; METRIC-QUALITY-04; METRIC-ROI-25; METRIC-ROI-24	forwarded requests during attribution degradation / eligible requests, plus observed continuity and latency impact	ratio, duration and status	eligible requests during degradation intervals
ASOR-SU-01	Attributed Energy Consumption per Workflow	sustainability	METRIC-SUSTAINABILITY-01; METRIC-SUSTAINABILITY-02; METRIC-ROI-44	compatible energy factor x matched workload units by workflow and region	kWh	matched workload units
ASOR-SU-02	Carbon Intensity per Accepted Result	sustainability	METRIC-SUSTAINABILITY-03; METRIC-ROI-45	compatible attributed kgCO2e / mature accepted outcomes	kgCO2e per accepted result	mature accepted outcomes
ASOR-SU-03	AI Energy Index Benchmark View	sustainability	METRIC-SUSTAINABILITY-06; METRIC-SUSTAINABILITY-07; METRIC-SUSTAINABILITY-09; METRIC-ROI-44; METRIC-ROI-45	qualified energy or emissions per comparable work unit / positive compatible reference intensity, retaining underlying intensity and factor provenance	indexed score and trend	positive compatible reference intensity for the same functional work unit

Evidence by view

ASOR-FND-01: Six-Layer Attribution Graph View

Decision: Inspect how an AI event connects owner, workflow, system, provider, cost and accepted outcome.

Read-only sources: OTel gen_ai metadata; workflow registry and run events; provider billing exports; read-only outcome receipts.

Customer context: tenant scope; workflow definition; acceptance rule.

Availability: Show only resolved edges and label missing layers; do not infer a complete graph from partial joins.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-QUALITY-03.six_layer_resolution_coverage; METRIC-QUALITY-03.exact_six_layer_resolution_share; METRIC-QUALITY-03.inferred_six_layer_resolution_share; METRIC-QUALITY-03.allocated_six_layer_resolution_share

Presentation contract: kind: attribution_graph; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; link_id; source_entity_id; target_entity_id; entity_layer; method; evidence_refs; source_revision; valid_from; valid_to; allocation_basis; scope rule: Project the six attribution layers and preserve exact, inferred, allocated and unresolved links separately. Coverage is distinct from independently labeled correctness; no label or link inherits another link’s state.

METRIC-QUALITY-03.six_layer_resolution_coverage

Property	Contract
Formula	Eligible events with all six layers resolved divided by all eligible events; preserve unresolved evidence.
Unit	ratio
Required inputs	usage; identity-map; org-map; feature-map; prices

METRIC-QUALITY-03.exact_six_layer_resolution_share

Property	Contract
Formula	Eligible fully resolved events in the declared exact resolution class divided by eligible events. Disclose overlap and classification precedence; completeness does not prove correctness.
Unit	ratio
Required inputs	usage; identity-map; org-map; feature-map; prices

METRIC-QUALITY-03.inferred_six_layer_resolution_share

Property	Contract
Formula	Eligible fully resolved events in the declared inferred resolution class divided by eligible events. Disclose overlap and classification precedence; completeness does not prove correctness.
Unit	ratio
Required inputs	usage; identity-map; org-map; feature-map; prices

METRIC-QUALITY-03.allocated_six_layer_resolution_share

Property	Contract
Formula	Eligible fully resolved events in the declared allocated resolution class divided by eligible events. Disclose overlap and classification precedence; completeness does not prove correctness.
Unit	ratio
Required inputs	usage; identity-map; org-map; feature-map; prices

ASOR-FND-02: Estate Coverage Ratio

Decision: Measure how much eligible AI activity and spend is represented in the record.

Read-only sources: provider billing exports; OTel and gateway metadata; SaaS administration metadata; workflow systems.

Customer context: covered-source inventory; eligibility boundary; source time window.

Availability: Compute each coverage ratio only against an independently supported finite positive eligible population. If total eligible spend or invocation volume is unknown, withhold that ratio and retain observed covered counts or spend plus the unknown-source inventory. Known uncovered sources reduce measured coverage; unknown or stale sources never become zero spend.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-COST-03.invoice_cost_coverage; METRIC-ADOPTION-01.managed_invocation_coverage; METRIC-ADOPTION-01.observed_managed_invocation_count; METRIC-ROI-37.covered_ai_spend

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Spend and invocation coverage use separate independently supported denominators. An unknown universe withholds that share while supported absolute counts and amounts remain available.

METRIC-COST-03.invoice_cost_coverage

Property	Contract
Formula	Attributed eligible cost divided by compatible eligible invoice cost; never force residuals into attributed owners.
Unit	ratio
Required inputs	usage; prices; invoices

METRIC-ADOPTION-01.managed_invocation_coverage

Property	Contract
Formula	Observed managed eligible invocations divided by an independently validated eligible invocation universe; discovered candidates alone do not establish that universe.
Unit	ratio
Required inputs	usage; discovery-evidence; egress-validation

METRIC-ADOPTION-01.observed_managed_invocation_count

Property	Contract
Formula	Count of observed managed eligible invocations in the declared population, independent of enterprise denominator evidence.
Unit	count
Required inputs	usage

METRIC-ROI-37.covered_ai_spend

Property	Contract
Formula	sum_unique(eligible_ai_cost_under_effective_covered_spend_contract)
Unit	USD
Required inputs	eligible_ai_cost_records; effective_covered_spend_contract; currency_and_cost_basis; trailing_period_and_tier_rules

ASOR-FND-03: Attribution Confidence Score

Decision: Distinguish exact, inferred and unresolved ownership; report shared-cost allocation separately.

Read-only sources: OTel correlation and identity metadata; identity and organization maps; workflow registry.

Customer context: confidence bands; identity map; effective-dated ownership mapping.

Availability: Exact and inferred links carry different labels; 0.95 is a cap, never a claim of certainty or precision.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: None declared.

Presentation contract: kind: ownership_link_records; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; link_id; asserted_owner_id; source_entity_id; c_oper; method; candidate_alternatives; evidence_refs; source_revision; valid_from; valid_to; scope rule: Present source-governed per-link c_oper, capped at 0.95, with method, alternatives and temporal validity. Do not aggregate confidence into correctness or invent a probability for unresolved ownership.

ASOR-FND-04: Unresolved Volume, Unknown Stays Unknown

Decision: Quantify the spend and traffic that cannot be resolved above the declared threshold.

Read-only sources: provider billing exports; OTel metadata; identity and workflow maps.

Customer context: resolution threshold; suspense bucket policy.

Availability: Route unresolved records to an explicit suspense state with diagnostic reason; never allocate by guess.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-COST-06.unresolved_owner_spend; METRIC-COST-06.unresolved_owner_spend_share; METRIC-ROI-39.unresolved_ownership_share; METRIC-ROI-39.unresolved_attribution_unit_count

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Retain independent unresolved counts, amounts and shares. Classify event or run grain explicitly; owner uncertainty cannot erase supported spend.

METRIC-COST-06.unresolved_owner_spend

Property	Contract
Formula	Eligible paid cost whose ownership remains unresolved; an incomplete mapping records unresolved ownership instead of inventing an owner.
Unit	USD
Required inputs	usage; prices; identity-map

METRIC-COST-06.unresolved_owner_spend_share

Property	Contract
Formula	unresolved_owner_spend divided by total eligible cost for the same tenant, population, window, currency and cost basis. Preserve the USD observation when the denominator is zero, unknown or incompatible; overlap among categories remains explicit.
Unit	ratio
Required inputs	usage; prices; identity-map

METRIC-ROI-39.unresolved_ownership_share

Property	Contract
Formula	eligible_units_in_unresolved_ownership_state / eligible_attribution_unit_count
Unit	ratio
Required inputs	ownership_state_records; eligible_attribution_unit_count; attribution_unit_definition; event_time_identity_scope

METRIC-ROI-39.unresolved_attribution_unit_count

Property	Contract
Formula	count_unique(eligible_attribution_units_in_unresolved_state)
Unit	attribution_unit
Required inputs	eligible_attribution_unit_ids; versioned_resolution_rule; source_population_completeness

ASOR-FND-05: Shared-Key and Machine-Account Disaggregation Rate

Decision: Assess whether shared principals can be traced to accountable workflows.

Read-only sources: provider key and principal metadata; OTel service identity; workflow and organization maps.

Customer context: shared-principal inventory; service-to-workflow mapping.

Availability: Report unresolved shared traffic separately from resolved traffic and preserve the original principal.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-39.shared_principal_workflow_resolution_share; METRIC-ROI-39.exact_ownership_share; METRIC-ROI-39.inferred_ownership_share; METRIC-ROI-39.unresolved_ownership_share

Presentation contract: kind: shared_principal_resolution; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; original_principal_id; accountable_workflow_id; method; evidence_refs; source_revision; valid_at; scope rule: Exact and eligible inferred workflow evidence can resolve shared principals. Allocation-only labels do not resolve ownership; retain exact, inferred and unresolved strata. Evaluate the three ownership shares on the same eligible shared-principal event cohort while retaining the original principal and unresolved records.

METRIC-ROI-39.shared_principal_workflow_resolution_share

Property	Contract
Formula	unique_eligible_shared_principal_events_with_exact_or_eligible_inferred_workflow / eligible_shared_principal_event_count
Unit	ratio
Required inputs	shared_principal_inventory; eligible_shared_principal_event_ids; workflow_resolution_evidence; versioned_resolution_rule

METRIC-ROI-39.exact_ownership_share

Property	Contract
Formula	eligible_units_in_exact_ownership_state / eligible_attribution_unit_count
Unit	ratio
Required inputs	ownership_state_records; eligible_attribution_unit_count; attribution_unit_definition; event_time_identity_scope

METRIC-ROI-39.inferred_ownership_share

Property	Contract
Formula	eligible_units_in_inferred_ownership_state / eligible_attribution_unit_count
Unit	ratio
Required inputs	ownership_state_records; eligible_attribution_unit_count; attribution_unit_definition; event_time_identity_scope

METRIC-ROI-39.unresolved_ownership_share

Property	Contract
Formula	eligible_units_in_unresolved_ownership_state / eligible_attribution_unit_count
Unit	ratio
Required inputs	ownership_state_records; eligible_attribution_unit_count; attribution_unit_definition; event_time_identity_scope

ASOR-FND-06: Source Telemetry Freshness and Lag View

Decision: Know whether source data is current enough for a financial or operational decision.

Read-only sources: collector health and pipeline telemetry; provider export timestamps; serving observability.

Customer context: freshness SLO; source schedule; staleness policy.

Availability: Show stale, missing, clock-skew and schema-drift states explicitly; negative intervals flag clock inconsistency and are not clipped to zero.

Freshness: required timestamps: source_event_at; observed_at; served_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Resolve each interval independently: source_to_observation_lag requires source_event_at and observed_at; ingest_to_served_lag requires observed_at and served_at; event_to_served_lag requires source_event_at and served_at. Missing served_at withholds only intervals that depend on it. Missing timestamps produce unavailable outputs with reasons; retain independently supported intervals. Never subtract served-dependent intervals to reconstruct source-to-observation lag. Staleness remains visible and negative intervals flag clock inconsistency without clipping.; timestamp mapping: observed_at is the first successful ingestion timestamp of the matched tenant/source/record revision; a versioned adapter mapping must establish this meaning. Last-seen, replay, query and data_as_of times cannot substitute.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-39.source_to_observation_lag; METRIC-QUALITY-04.ingest_to_served_lag; METRIC-QUALITY-04.event_to_served_lag

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Each lag depends only on its named timestamp endpoints for the same source record and revision. Missing served_at cannot withhold source-to-observation lag or manufacture it through subtraction.

METRIC-ROI-39.source_to_observation_lag

Property	Contract
Formula	observed_at - source_event_at
Required timestamps	source_event_at; observed_at

METRIC-QUALITY-04.ingest_to_served_lag

Property	Contract
Formula	served_at - observed_at
Required timestamps	observed_at; served_at

METRIC-QUALITY-04.event_to_served_lag

Property	Contract
Formula	served_at - source_event_at
Required timestamps	source_event_at; served_at

ASOR-FND-07: Owner Correction and Override Log

Decision: Replay who corrected an attribution and what historical values changed.

Read-only sources: customer-approved correction records; immutable audit records; effective-dated identity maps.

Customer context: correction authority; effective date; reason code.

Availability: Expose original and restated values; never destructively overwrite a prior close.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: None declared.

Presentation contract: kind: correction_log; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; correction_id; asserted_mapping; source_provenance; effective_at; transaction_at; supersedes; source_revision; reason; scope rule: Project append-only corrections with effective and transaction time. Supersession preserves the prior assertion; history is not rewritten into a current-state-only list.

ASOR-FND-08: Undeclared or Shadow AI Spend Discovery

Decision: Find AI spend outside governed inventories without claiming individual surveillance.

Read-only sources: corporate card and marketplace exports; provider billing exports; network or egress metadata; approved discovery evidence.

Customer context: governed channel inventory; aggregate ownership boundary.

Availability: Report candidate channels and economic context; individual attribution requires separate authorized evidence.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-SECURITY-01.attributed_channel_cost

Presentation contract: kind: discovery_inventory; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; candidate_id; entity_id; channel_classification; validation_state; owner_state; evidence_refs; source_revision; observed_at; scope rule: Only positively validated unmanaged channels enter the observed unmanaged inventory. Retain unvalidated candidates separately; missing price does not hide supported entities.

METRIC-SECURITY-01.attributed_channel_cost

Property	Contract
Formula	Matched eligible paid cost for detected channels with compatible usage and rates; missing prices withhold cost while preserving the observed count.
Unit	USD
Required inputs	discovery-evidence; usage; prices

ASOR-WE-01: Cost per Accepted Result

Decision: Compare the cost of delivering accepted work across workflow alternatives.

Read-only sources: provider billing exports; OTel usage and duration metadata; workflow runs and retries; read-only outcome receipts; human review metadata when available.

Customer context: named workflow; cost boundary; acceptance rule; maturity window.

Availability: Include paid failures, retries and supported tool cost; withhold if accepted outcomes or complete cost basis is unavailable.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Evaluate each named output independently within its tenant, compatible cohort, period, unit, evidence basis and source revision.

METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Property	Contract
Formula	full_workflow_cohort_cost / unique_mature_accepted_outcome_count
Unit	currency/accepted_outcome
Required inputs	full_workflow_cohort_cost; unique_mature_accepted_outcome_count; cost_component_completeness; cost_allocation_basis; cohort_acceptance_and_maturity_contract

ASOR-WE-02: Gross Cost per Invocation vs Effective Cost per Accepted Result

Decision: Explain why a low call price may still produce expensive accepted work.

Read-only sources: provider billing exports; OTel token and duration metadata; workflow retries and outcomes.

Customer context: workflow cohort; accepted-result rule; cost inclusions.

Availability: Keep invocation and accepted-result populations separate and expose failed or rejected work.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-01.full_workflow_cost_per_accepted_outcome; METRIC-ROI-23.gross_cost_per_invocation

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Pair distinct accepted-outcome and model-invocation grains. A conversation turn is not an invocation; missing one denominator withholds only its own result.

METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Property	Contract
Formula	full_workflow_cohort_cost / unique_mature_accepted_outcome_count
Unit	currency/accepted_outcome
Required inputs	full_workflow_cohort_cost; unique_mature_accepted_outcome_count; cost_component_completeness; cost_allocation_basis; cohort_acceptance_and_maturity_contract

METRIC-ROI-23.gross_cost_per_invocation

Property	Contract
Formula	complete_supported_invocation_cost / eligible_invocation_count
Unit	USD_per_invocation
Required inputs	invocation_identity_and_lifecycle; complete_supported_invocation_cost; eligible_invocation_count; currency_and_cost_basis

ASOR-WE-03: Fully Loaded Workflow Cost Breakdown

Decision: Identify which cost components drive the cost of accepted work.

Read-only sources: provider billing exports; cloud cost exports; OTel usage metadata; workflow/tool events; customer labor records when available.

Customer context: included cost components; allocation method; labor evidence boundary.

Availability: Label fully loaded only when every applicable component is evidenced and deduplicated; show supported partial costs with their missing components when the full boundary is unavailable.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-01.full_workflow_cohort_cost

Presentation contract: kind: cost_component_ledger; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; cohort_id; component_id; cost_category; amount; currency; cost_basis; source_record_ids; allocation_basis; coverage; source_revision; scope rule: Components form a disjoint ledger reconciled to the full workflow cohort cost. Preserve paid failures and retries; human, provider and tool costs cannot overlap.

METRIC-ROI-01.full_workflow_cohort_cost

Property	Contract
Formula	sum(billable_leaf_costs) + sum(human_review_rework_costs) + sum(disjoint_support_allocations), after unique billable-leaf deduplication and cost-boundary reconciliation
Unit	currency
Required inputs	billable_leaf_costs; human_review_rework_costs; disjoint_support_allocations; full_cost_boundary; cost_component_completeness

ASOR-WE-04: Net Realizable Benefit Bridge

Decision: Evaluate whether a change creates realizable economic benefit after customer-borne costs.

Read-only sources: workflow cost ledger; customer finance inputs; accepted outcome evidence; implementation and labor records.

Customer context: benefit definition; subscription and implementation fees; period; decision baseline.

Availability: Require attributable incremental benefit, a compatible baseline, complete investment and no duplicate costs or benefits; preserve negative net benefit and withhold ROI for a zero or unknown investment denominator.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-02.incremental_financial_benefit_before_cost; METRIC-ROI-02.complete_incremental_investment; METRIC-ROI-02.incremental_net_benefit

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Evaluate each named output independently within its tenant, compatible cohort, period, unit, evidence basis and source revision.

METRIC-ROI-02.incremental_financial_benefit_before_cost

Property	Contract
Formula	sum(entry.amount_before_included_cost for unique, realized entries in realized_benefit_entries under benefit_cost_inclusion_map)
Unit	currency
Required inputs	realized_benefit_entries; benefit_cost_inclusion_map; financial_realization_evidence; comparison_contract

METRIC-ROI-02.complete_incremental_investment

Property	Contract
Formula	sum(unique incrementally incurred amounts in incremental_cost_components under cost_inclusion_map)
Unit	currency
Required inputs	investment_scope; incremental_cost_components; cost_completeness_evidence; cost_inclusion_map

METRIC-ROI-02.incremental_net_benefit

Property	Contract
Formula	if net_benefit_evidence_mode = gross: incremental_financial_benefit_before_cost - complete_incremental_investment; if net_benefit_evidence_mode = independently_complete_already_net: evidenced_already_net_change
Unit	currency
Required inputs	net_benefit_evidence_mode; comparison_contract; benefit_cost_inclusion_map

ASOR-WE-05: Cash Savings vs Released Capacity Split

Decision: Prevent double counting while separating cash reduction from capacity release.

Read-only sources: finance actuals; payroll or time records; workflow outcomes; accepted throughput.

Customer context: cash boundary; labor basis; reference hours; redeployment evidence.

Availability: Match task mix, quality, period, roles, labor basis and acceptance rules. Include failed work, review, correction, rework and disjoint overhead once. Preserve signed net work equivalence. Newly enabled demand is work equivalence, not automatic hours released. Claim AI-attributable release only with the output-specific identified counterfactual and demand boundary; redeployment requires independent evidence. Unsupported baselines withhold dependent capacity outputs; hours never imply payroll savings.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-03.net_realized_cash_expense_reduction; METRIC-ROI-04.net_labor_equivalent_hours; METRIC-ROI-04.ai_attributable_released_hours; METRIC-ROI-04.evidenced_redeployed_hours

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Cash reduction, estimated labor-equivalent capacity, attributable released time and evidenced redeployment remain separate. Capacity is never automatically payroll savings.

METRIC-ROI-03.net_realized_cash_expense_reduction

Property	Contract
Formula	gross_cash_expense_reduction - disjoint_incremental_implementation_and_operating_cash_cost
Unit	currency
Required inputs	gross_cash_expense_reduction; disjoint_incremental_implementation_and_operating_cash_cost; benefit_cost_inclusion_map

METRIC-ROI-04.net_labor_equivalent_hours

Property	Contract
Formula	baseline_equivalent_hours - all_current_human_hours - disjoint_setup_maintenance_and_other_overhead_hours
Unit	labor_hour
Required inputs	baseline_equivalent_hours; all_current_human_hours; current_effort_evidence_basis; disjoint_setup_maintenance_and_other_overhead_hours; role_and_labor_basis; overhead_inclusion_map

METRIC-ROI-04.ai_attributable_released_hours

Property	Contract
Formula	identified_expected_human_hours_without_intervention - identified_expected_human_hours_with_intervention_for_existing_demand
Unit	labor_hour
Required inputs	existing_demand_evidence; complete_cohort_effort; identified_comparison; task_mix_and_labor_basis; effect_uncertainty

METRIC-ROI-04.evidenced_redeployed_hours

Property	Contract
Formula	sum(nonoverlapping_role_hours_linked_to_verified_redeployment_assignments_and_accepted_work)
Unit	labor_hour
Required inputs	redeployment_assignments; role_effort_records; destination_accepted_work; source_and_destination_overlap_ledger

ASOR-WE-06: Avoidable Workflow Cost

Decision: Locate waste in failed trajectories, retries, rejected drafts and duplicate work.

Read-only sources: OTel spans; workflow run and attempt events; provider billing exports; outcome receipts.

Customer context: terminal-state rules; duplicate definition; retry inclusion policy.

Availability: Keep all paid attempts in total cost. Report failure cost separately from demonstrated avoidable cost; zero accepted results is a valid cohort with unavailable cost per accepted result.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-QUALITY-05.team_failed_attempt_spend; METRIC-ROI-14.abandoned_work_cost; METRIC-ROI-14.duplicate_work_cost; METRIC-ROI-14.unnecessary_retry_cost; METRIC-ROI-14.avoidable_workflow_cost_scenario

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Paid work may have overlapping failure classifications; deduplicate before any total. Classified waste is not automatically avoidable and a prevention scenario must preserve comparable accepted work.

METRIC-QUALITY-05.team_failed_attempt_spend

Property	Contract
Formula	Eligible paid cost of failed attempts by effective owner, including retry-rescued attempts.
Unit	USD
Required inputs	usage; prices; identity-map; org-map; workflow-events

METRIC-ROI-14.abandoned_work_cost

Property	Contract
Formula	sum(unique_paid_costs_classified_abandoned)
Unit	currency
Required inputs	billable_attempt_lineage; abandonment_and_maturity_evidence; disjoint_waste_classification

METRIC-ROI-14.duplicate_work_cost

Property	Contract
Formula	sum(unique_paid_costs_classified_duplicate)
Unit	currency
Required inputs	billable_attempt_lineage; duplicate_business_work_evidence; disjoint_waste_classification

METRIC-ROI-14.unnecessary_retry_cost

Property	Contract
Formula	sum(unique_paid_costs_classified_unnecessary_retry)
Unit	currency
Required inputs	billable_attempt_lineage; retry_necessity_evidence; disjoint_waste_classification

METRIC-ROI-14.avoidable_workflow_cost_scenario

Property	Contract
Formula	complete_observed_workflow_cost - supported_prevention_scenario_cost - disjoint_prevention_cost
Unit	USD
Required inputs	complete_observed_workflow_cost; supported_prevention_scenario_cost; disjoint_prevention_cost; comparable_accepted_work_evidence; versioned_prevention_scenario

ASOR-WE-07: Budget Owner Chargeback and Showback Ledger

Decision: Allocate covered AI spend to P&L owners with evidence boundaries.

Read-only sources: provider and cloud invoices; cost-center and organization maps; workflow ownership metadata.

Customer context: cost-center hierarchy; shared allocation rule; chargeback eligibility threshold.

Availability: Reconcile to invoices with a signed residual; low-confidence lines remain showback or suspense.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-37.direct_cost_amount; METRIC-ROI-37.shared_allocated_cost_amount; METRIC-ROI-37.unallocated_cost_amount; METRIC-ROI-37.invoice_reconciliation_signed_residual

Presentation contract: kind: owner_cost_ledger; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; owner_id; cost_record_id; direct_or_shared_or_unallocated; allocation_rule_version; amount; currency; cost_basis; source_revision; scope rule: Reconcile disjoint direct, shared-allocated and unallocated amounts with the invoice residual. Allocation-only ownership remains labeled and does not imply attribution correctness.

METRIC-ROI-37.direct_cost_amount

Property	Contract
Formula	sum(unique_direct_cost_components)
Unit	currency
Required inputs	direct_cost_records; source_pool_id; cost_basis

METRIC-ROI-37.shared_allocated_cost_amount

Property	Contract
Formula	sum(unique_shared_cost_allocations)
Unit	currency
Required inputs	shared_cost_pool; allocation_records; allocation_policy_version

METRIC-ROI-37.unallocated_cost_amount

Property	Contract
Formula	sum(source_cost_components_remaining_unallocated)
Unit	currency
Required inputs	source_cost_pool; direct_assignment_ledger; shared_allocation_ledger; unallocated_component_records

METRIC-ROI-37.invoice_reconciliation_signed_residual

Property	Contract
Formula	recorded_reconciled_cost - authoritative_invoice_cost
Unit	currency
Required inputs	recorded_reconciled_cost; authoritative_invoice_cost; invoice_scope_and_adjustment_map

ASOR-WE-08: Recurring Reconstruction Effort Saved

Decision: Measure the operating effort removed from monthly attribution close.

Read-only sources: time records; close workflow events; report generation and exception-review logs.

Customer context: baseline period; role rates; close scope; comparability rule.

Availability: Require comparable periods and preserve loaded cost as a scenario unless actual cash expense changed.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-40.finops_capacity_valuation_scenario; METRIC-ROI-40.finops_process_effort_saved

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Saved person-hours equal comparable baseline minus current effort, preserving negative differences. Valued capacity is a scenario and remains separate from realized cash.

METRIC-ROI-40.finops_capacity_valuation_scenario

Property	Contract
Formula	supported_finops_released_hours * declared_role_labor_value_rate
Unit	currency
Required inputs	supported_finops_released_hours; declared_role_labor_value_rate; valuation_basis; benefit_overlap_map

METRIC-ROI-40.finops_process_effort_saved

Property	Contract
Formula	baseline_comparable_process_person_hours - current_process_person_hours
Unit	person_hour
Required inputs	baseline_comparable_process_person_hours; current_process_person_hours; matched_process_role_period_and_effort_basis

ASOR-WE-09: Workflow Economics Decision Brief View

Decision: Create a repeatable scale, change or stop recommendation.

Read-only sources: all approved workflow economics sources.

Customer context: decision owner; alternative or scenario; acceptance rule; review date.

Availability: A brief is publishable only when every dependent output has an evidence state and missing inputs are disclosed.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-01.full_workflow_cost_per_accepted_outcome; METRIC-ROI-02.incremental_net_benefit; METRIC-ROI-02.financial_roi; METRIC-ROI-12.marginal_return_on_additional_budget; METRIC-ROI-33.realized_cash_payback_time; METRIC-ROI-33.projected_cash_payback_time

Presentation contract: kind: decision_brief; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; decision_id; workflow_id; recommendation; scope; assumptions; alternatives; evidence_refs; method_version; source_revision; scope rule: A decision brief binds each recommendation to independently available evidence, alternatives and assumptions. Realized and projected payback remain distinct; unavailable ROI cannot become zero.

METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Property	Contract
Formula	full_workflow_cohort_cost / unique_mature_accepted_outcome_count
Unit	currency/accepted_outcome
Required inputs	full_workflow_cohort_cost; unique_mature_accepted_outcome_count; cost_component_completeness; cost_allocation_basis; cohort_acceptance_and_maturity_contract

METRIC-ROI-02.incremental_net_benefit

Property	Contract
Formula	if net_benefit_evidence_mode = gross: incremental_financial_benefit_before_cost - complete_incremental_investment; if net_benefit_evidence_mode = independently_complete_already_net: evidenced_already_net_change
Unit	currency
Required inputs	net_benefit_evidence_mode; comparison_contract; benefit_cost_inclusion_map

METRIC-ROI-02.financial_roi

Property	Contract
Formula	incremental_net_benefit / complete_incremental_investment
Unit	ratio
Required inputs	incremental_net_benefit; complete_incremental_investment; investment_scope; benefit_cost_inclusion_map

METRIC-ROI-12.marginal_return_on_additional_budget

Property	Contract
Formula	incremental_budget_net_benefit_change / incremental_budget_cost_change
Unit	ratio
Required inputs	incremental_budget_net_benefit_change; incremental_budget_cost_change; identified_budget_comparison

METRIC-ROI-33.realized_cash_payback_time

Property	Contract
Formula	earliest_t_with(cumulative_cash_benefits(t) >= cumulative_cash_costs(t)) - investment_start_time, after a positive cash outlay
Unit	day
Required inputs	dated_cash_benefit_ledger; dated_cash_cost_ledger; positive_initial_cash_outlay; investment_start_time; cash_scope

METRIC-ROI-33.projected_cash_payback_time

Property	Contract
Formula	earliest_projected_t_with(projected_cumulative_cash_benefits(t) >= projected_cumulative_cash_costs(t)) - investment_start_time
Unit	day
Required inputs	dated_observed_cash_flows; projected_cash_flow_assumptions; positive_investment_scope; forecast_horizon

ASOR-WE-10: Covered AI Spend Tier Tracker

Decision: Know which licensed spend band and decision scope the record covers.

Read-only sources: provider billing exports; contract and commitment metadata; tenant entitlement records.

Customer context: licensed tier; covered providers; period.

Availability: This is coverage and contract context, not a customer ROI result or a claim that all enterprise spend is covered.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-37.covered_ai_spend

Presentation contract: kind: contract_tier_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; contract_id; contract_version; tier_id; effective_from; effective_to; trailing_period; currency; cost_basis; source_revision; scope rule: Covered spend uses the effective contract and declared trailing period. An unsupported contract tier remains unknown independently of supported spend.

METRIC-ROI-37.covered_ai_spend

Property	Contract
Formula	sum_unique(eligible_ai_cost_under_effective_covered_spend_contract)
Unit	USD
Required inputs	eligible_ai_cost_records; effective_covered_spend_contract; currency_and_cost_basis; trailing_period_and_tier_rules

ASOR-AD-01: Workflow-Level Adoption Funnel

Decision: Measure progression from experimentation to repeatable embedded production work.

Read-only sources: workflow registry; usage and outcome events; product event metadata.

Customer context: stage definitions; cohort window; approved workflow inventory.

Availability: Aggregate at team or workflow level; never rank individual employees.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-26.workflow_stage_count; METRIC-ROI-26.workflow_stage_conversion_rate

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Declare exclusive current stage or cumulative funnel policy. Conversion uses a frozen mature source-stage cohort and target transition evidence, not the current stage count as an assumed denominator.

METRIC-ROI-26.workflow_stage_count

Property	Contract
Formula	count_unique(eligible_approved_workflow_definitions_in_declared_stage)
Unit	workflow
Required inputs	approved_workflow_registry; stage_evidence; versioned_stage_definition; cohort_window_and_stage_policy

METRIC-ROI-26.workflow_stage_conversion_rate

Property	Contract
Formula	unique_mature_source_stage_workflows_reaching_target_stage / eligible_mature_source_stage_workflow_count
Unit	ratio
Required inputs	source_stage_cohort_ids; target_stage_transition_evidence; eligible_mature_source_stage_workflow_count; versioned_stage_and_maturity_rules

ASOR-AD-02: Licensed Seat vs Active Workflow Utilization

Decision: Find shelfware and compare licensed access with outcome-producing use.

Read-only sources: SaaS administration exports; workflow and outcome events; license records.

Customer context: license scope; active-use rule; seat-to-workflow mapping.

Availability: Active use requires observed workflow activity; a login or exposed UI is not an accepted outcome.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-20.outcome_producing_license_utilization_rate

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Evaluate licensed seat and licensed workflow entity types separately against the effective complete roster and mature accepted work.

METRIC-ROI-20.outcome_producing_license_utilization_rate

Property	Contract
Formula	distinct_eligible_licensed_entities_with_mature_accepted_work / eligible_licensed_entity_count
Unit	ratio
Required inputs	license_roster_and_effective_terms; licensed_entity_to_work_evidence; mature_accepted_work_records; eligible_licensed_entity_count; licensed_entity_type

ASOR-AD-03: Accepted-Work Adoption Rate

Decision: Measure how much target work is completed with AI assistance and accepted.

Read-only sources: workflow systems; AI usage metadata; outcome receipts.

Customer context: target task population; AI-assisted definition; acceptance rule.

Availability: Require comparable task volume and mature acceptance; do not substitute usage volume for accepted work.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-26.productive_workflow_adoption_rate

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Use the versioned mapping from task or workflow instances to eligible target work, with supported AI assistance and mature acceptance.

METRIC-ROI-26.productive_workflow_adoption_rate

Property	Contract
Formula	accepted_ai_assisted_workflow_instance_count / eligible_workflow_instance_count
Unit	ratio
Required inputs	accepted_ai_assisted_workflow_instance_count; eligible_workflow_instance_count; workflow_eligibility_contract; supported_exposure_scope

ASOR-AD-04: Human-in-the-Loop Override and Abandonment Rate

Decision: Identify when users discard, heavily edit or bypass AI work.

Read-only sources: existing output delivery, revision, review, discard, revert and terminal-state metadata; source-native aggregate union counts with established population and overlap.

Customer context: versioned output decision grain and eligibility; override and heavy-edit definitions, measurement method and threshold; observation horizon and late-arrival allowance; terminal disposition and resumption mapping.

Availability: Return unavailable with a specific reason for missing required evidence, incomplete terminal classification, unknown category overlap or a zero denominator; retain immature windows and supported counts. Inactivity is not abandonment, missing events are not zero override, and external benchmarks cannot supply dispositions.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-09.ai_output_override_or_abandonment_rate

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Evaluate each named output independently within its tenant, compatible cohort, period, unit, evidence basis and source revision.

METRIC-ROI-09.ai_output_override_or_abandonment_rate

Property	Contract
Formula	count_unique(eligible_mature_terminal_output_units_with_qualifying_override_or_abandonment) / count_unique(eligible_mature_terminal_output_units)
Unit	ratio
Required inputs	output_population_and_grain_contract; versioned_override_and_abandonment_policy; observation_maturity_and_source_coverage; result_source_provenance

ASOR-AD-05: Cross-Departmental Capability Diffusion Map

Decision: See where accepted workflows are concentrated and where enablement lags.

Read-only sources: organization and role maps; workflow and outcome events; usage metadata.

Customer context: aggregate organizational dimensions; geography policy; cohort window.

Availability: Use minimum cohort sizes and aggregate groups; no individual surveillance or stack ranking.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-01.full_workflow_cost_per_accepted_outcome; METRIC-ROI-27.cohort_accepted_unit_cost_change; METRIC-ROI-26.productive_workflow_adoption_rate; METRIC-ROI-26.accepted_ai_assisted_workflow_instance_count

Presentation contract: kind: aggregate_capability_map; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: tenant_id; business_unit; role_group; geography; cohort_id; period; aggregation_authorization; source_revision; scope rule: Show authorized aggregate groups. Accepted-instance counts remain available without prices or an eligible-universe denominator; incompatible work units and small restricted groups are never pooled.

METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Property	Contract
Formula	full_workflow_cohort_cost / unique_mature_accepted_outcome_count
Unit	currency/accepted_outcome
Required inputs	full_workflow_cohort_cost; unique_mature_accepted_outcome_count; cost_component_completeness; cost_allocation_basis; cohort_acceptance_and_maturity_contract

METRIC-ROI-27.cohort_accepted_unit_cost_change

Property	Contract
Formula	current_cohort_cost_per_accepted_outcome - baseline_cohort_cost_per_accepted_outcome
Unit	currency/accepted_outcome
Required inputs	cohort_metadata; current_cohort_cost_per_accepted_outcome; baseline_cohort_cost_per_accepted_outcome; pre_treatment_comparable_task_contract

METRIC-ROI-26.productive_workflow_adoption_rate

Property	Contract
Formula	accepted_ai_assisted_workflow_instance_count / eligible_workflow_instance_count
Unit	ratio
Required inputs	accepted_ai_assisted_workflow_instance_count; eligible_workflow_instance_count; workflow_eligibility_contract; supported_exposure_scope

METRIC-ROI-26.accepted_ai_assisted_workflow_instance_count

Property	Contract
Formula	count_unique(eligible_mature_accepted_ai_assisted_workflow_instances)
Unit	workflow_instance
Required inputs	workflow_instance_identity; supported_ai_assistance_evidence; mature_unrevoked_acceptance; declared_reporting_cohort

ASOR-AD-06: Program Mandate vs Realized Production Pull

Decision: Compare executive targets with recurring accepted production work.

Read-only sources: program target records; workflow registry; usage and accepted outcome events.

Customer context: mandate target; production qualification rule; review period.

Availability: Targets and realized production are separate evidence classes; a mandate does not establish adoption.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-30.recurring_accepted_program_target_attainment

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Use a frozen program target with compatible workflow-instance unit and period. Target attainment is distinct from a causal baseline or observed estate-wide adoption.

METRIC-ROI-30.recurring_accepted_program_target_attainment

Property	Contract
Formula	verified_recurring_accepted_workflow_volume / frozen_declared_program_target_volume
Unit	ratio
Required inputs	verified_recurring_accepted_workflow_volume; frozen_declared_program_target_volume; versioned_production_and_recurrence_rules; matched_program_work_unit_and_period

ASOR-Q-01: End-to-End Task Acceptance Rate

Decision: Measure whether AI-initiated work reaches its intended terminal state.

Read-only sources: workflow run events; downstream system-of-record status metadata; acceptance receipts.

Customer context: terminal accepted state; maturity window; eligible workflow cohort.

Availability: Completion alone is not acceptance; pending and revoked outcomes remain distinct.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-QUALITY-06.outcome_acceptance_rate

Presentation contract: kind: scalar_projection; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Evaluate each named output independently within its tenant, compatible cohort, period, unit, evidence basis and source revision.

METRIC-QUALITY-06.outcome_acceptance_rate

Property	Contract
Formula	Distinct eligible completed/mature runs with unrevoked customer acceptance / eligible runs under the declared cohort rule. Basic acceptance evidence belongs to Core; customer semantics/version and absence/rejection distinction are mandatory.
Unit	ratio
Required inputs	workflow-registry; workflow-events; outcomes

ASOR-Q-02: Retry, Fallback and Loop Amplification Factor

Decision: Detect runaway trajectories and explain extra cost.

Read-only sources: OTel spans; workflow attempt events; routing and tool-call metadata.

Customer context: nominal path; retry and fallback classification; run boundary.

Availability: Show observed amplification and its cost; do not infer a nominal path when it is undeclared.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-01.full_workflow_cohort_cost; METRIC-ROI-16.evaluated_redundant_step_cost; METRIC-ROI-16.nominal_path_call_amplification; METRIC-ROI-16.observed_amplification_call_cost

Presentation contract: kind: factor_and_cost; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Observed total workflow cost is distinct from incremental amplification cost. Evaluated redundant-step cost requires comparable reduced-path evidence. Model and tool call grain excludes duplicate nested spans. Observed amplification-call cost includes only actual billable calls positively aligned as additional to the nominal path. Useful rescue costs stay included and the amount is not automatically avoidable.

METRIC-ROI-01.full_workflow_cohort_cost

Property	Contract
Formula	sum(billable_leaf_costs) + sum(human_review_rework_costs) + sum(disjoint_support_allocations), after unique billable-leaf deduplication and cost-boundary reconciliation
Unit	currency
Required inputs	billable_leaf_costs; human_review_rework_costs; disjoint_support_allocations; full_cost_boundary; cost_component_completeness

METRIC-ROI-16.evaluated_redundant_step_cost

Property	Contract
Formula	complete_cost_of_evaluated_original_path - complete_cost_of_evaluated_qualified_reduced_path
Unit	currency
Required inputs	readable_variant_or_ablation_results; original_and_reduced_path_complete_costs; acceptance_quality_and_constraint_parity; comparison_contract

METRIC-ROI-16.nominal_path_call_amplification

Property	Contract
Formula	observed_model_and_tool_call_count / declared_nominal_path_call_count
Unit	factor
Required inputs	complete_run_attempt_call_metadata; observed_model_and_tool_call_count; declared_nominal_path_call_count; versioned_workflow_and_call_grain

METRIC-ROI-16.observed_amplification_call_cost

Property	Contract
Formula	sum_unique(supported_billable_leaf_cost_of_calls_aligned_as_additional_to_nominal_path)
Unit	USD
Required inputs	actual_to_nominal_call_path_mapping; versioned_nominal_path; unique_billable_leaf_cost_records; compatible_currency_cost_basis_and_run_scope; retry_fallback_loop_classification

ASOR-Q-03: Human Review and Correction Burden

Decision: Quantify human effort required before acceptance.

Read-only sources: review workflow events; ticket or document revision metadata; time records when available.

Customer context: review boundary; correction threshold; labor evidence source.

Availability: Missing review evidence is unavailable, never zero minutes.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-29.downstream_review_hours; METRIC-ROI-29.downstream_remediation_hours; METRIC-ROI-09.human_review_correction_minutes_per_accepted_output

Presentation contract: kind: effort_per_accepted_output; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Union overlapping intervals separately per person, then sum person-time across people. All failed and rejected attempts remain in the same cohort effort numerator; calibrated effort remains distinct from recorded time.

METRIC-ROI-29.downstream_review_hours

Property	Contract
Formula	sum(nonoverlapping_recorded_downstream_review_person_hours)
Unit	labor_hour
Required inputs	downstream_review_records; recorded_person_time; workflow_handoff_lineage

METRIC-ROI-29.downstream_remediation_hours

Property	Contract
Formula	sum(nonoverlapping_recorded_downstream_remediation_person_hours)
Unit	labor_hour
Required inputs	downstream_defect_reopen_records; recorded_remediation_person_time; workflow_handoff_lineage

METRIC-ROI-09.human_review_correction_minutes_per_accepted_output

Property	Contract
Formula	complete_nonoverlapping_review_and_correction_person_minutes / mature_accepted_output_count
Unit	person_minute_per_accepted_output
Required inputs	review_and_correction_effort_evidence; complete_nonoverlapping_review_and_correction_person_minutes; mature_accepted_output_count; matched_output_cohort_and_effort_basis

ASOR-Q-04: Model and Provider Swap Impact Matrix

Decision: Compare cost and outcome quality before changing a route.

Read-only sources: provider billing exports; OTel model and token metadata; workflow outcomes; routing events.

Customer context: candidate routes; matching rules; quality floor; period.

Availability: Price-only scenarios cannot be presented as quality-preserving alternatives without matched accepted outcomes.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-13.current_route_cost_per_accepted_outcome; METRIC-QUALITY-06.outcome_acceptance_rate

Presentation contract: kind: route_comparison_matrix; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: route_id; comparison_group_id; matched_cohort_id; model_provider_version; source_revision; acceptance_rule_version; scope rule: Evaluate each paired route under the same versioned work, quality and maturity cohort. Keep price-only scenarios separately labeled; route evidence cannot inherit another route’s acceptance.

METRIC-ROI-13.current_route_cost_per_accepted_outcome

Property	Contract
Formula	current_route_complete_cost / current_route_mature_accepted_outcomes
Unit	currency/accepted_outcome
Required inputs	current_route_complete_cost; current_route_mature_accepted_outcomes; route_contract; cohort_acceptance_and_maturity_contract

METRIC-QUALITY-06.outcome_acceptance_rate

Property	Contract
Formula	Distinct eligible completed/mature runs with unrevoked customer acceptance / eligible runs under the declared cohort rule. Basic acceptance evidence belongs to Core; customer semantics/version and absence/rejection distinction are mandatory.
Unit	ratio
Required inputs	workflow-registry; workflow-events; outcomes

ASOR-Q-05: Drift and Outcome Degradation Alert View

Decision: Identify material drops in accepted work after a model, prompt or data change.

Read-only sources: model and prompt version metadata; workflow outcomes; usage and cost records.

Customer context: change-point rule; minimum cohort; quality floor.

Availability: Alert only when the comparison cohort and maturity are sufficient; availability is separate from notification delivery.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-QUALITY-06.outcome_acceptance_rate; METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Presentation contract: kind: trend_and_alert_records; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: change_point_id; cohort_version; alert_id; generated_at; delivery_state; evidence_refs; scope rule: Repeated versioned results support trends under compatible periods. Alert generation and delivered notification remain distinct.

METRIC-QUALITY-06.outcome_acceptance_rate

Property	Contract
Formula	Distinct eligible completed/mature runs with unrevoked customer acceptance / eligible runs under the declared cohort rule. Basic acceptance evidence belongs to Core; customer semantics/version and absence/rejection distinction are mandatory.
Unit	ratio
Required inputs	workflow-registry; workflow-events; outcomes

METRIC-ROI-01.full_workflow_cost_per_accepted_outcome

Property	Contract
Formula	full_workflow_cohort_cost / unique_mature_accepted_outcome_count
Unit	currency/accepted_outcome
Required inputs	full_workflow_cohort_cost; unique_mature_accepted_outcome_count; cost_component_completeness; cost_allocation_basis; cohort_acceptance_and_maturity_contract

ASOR-SG-01: Decision Provenance Reconstruction View

Decision: Reconstruct who invoked an AI workflow, what ran, what records changed and who accepted it.

Read-only sources: identity and organization maps; OTel metadata; workflow and decision events; downstream record metadata.

Customer context: decision scope; verifier rule; retention period.

Availability: Missing edges remain explicit and lower lineage completeness; no prompt or completion payload is required.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-SECURITY-07.work_cost_lineage_coverage; METRIC-SECURITY-07.work_evidence_decision_lineage_coverage

Presentation contract: kind: decision_lineage; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: actor_id; workflow_id; model_provider_id; cost_reference; policy_revision; decision_id; acceptance_reference; edge_provenance; valid_at; missing_edges; scope rule: Coverage retains its governed cost-weighted denominator; typed lineage cannot infer absent edges from a coverage percentage.

METRIC-SECURITY-07.work_cost_lineage_coverage

Property	Contract
Formula	Eligible scoped cost traceable to registered work and actual lifecycle evidence divided by eligible scoped cost; acceptance and decisions are not required for this partial lineage.
Unit	ratio
Required inputs	usage; prices; workflow-registry; workflow-events

METRIC-SECURITY-07.work_evidence_decision_lineage_coverage

Property	Contract
Formula	Eligible cost linked through work, valid evidence and an actual versioned decision divided by eligible scoped cost.
Unit	ratio
Required inputs	usage; prices; workflow-registry; workflow-events; outcomes; decision-events

ASOR-SG-02: Shape Advisory and Gate Policy Status

Decision: See advisory recommendations and explicitly opted-in customer enforcement state.

Read-only sources: policy events; serving observability; customer policy configuration.

Customer context: workload policy; opt-in gate state; availability SLO.

Availability: Shape is advisory and always forwards; Gate is per-workload customer opt-in. No blocking-enforcement claim is implied.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-SECURITY-06.policy_evaluation_count; METRIC-SECURITY-06.advisory_breach_count; METRIC-SECURITY-06.evaluation_availability; METRIC-SECURITY-06.evaluation_gap_duration

Presentation contract: kind: policy_status_and_counts; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: workload_id; policy_revision; shape_mode; gate_opt_in_state; signed_opt_in_reference; effective_at; source_revision; scope rule: Original opt-in state and advisory forwarding are typed evidence. Configuration does not establish enforcement success; Shape remains the default and Gate is customer opt-in.

METRIC-SECURITY-06.policy_evaluation_count

Property	Contract
Formula	Observed eligible advisory policy evaluations in the declared window.
Unit	count
Required inputs	policy-events

METRIC-SECURITY-06.advisory_breach_count

Property	Contract
Formula	Observed advisory breaches under the effective policy; this does not establish policy availability during an observation gap.
Unit	count
Required inputs	policy-events

METRIC-SECURITY-06.evaluation_availability

Property	Contract
Formula	Available and current evaluations over the declared eligible observed path, with policy and serving coverage reconciled.
Unit	ratio
Required inputs	policy-events; serving-observability

METRIC-SECURITY-06.evaluation_gap_duration

Property	Contract
Formula	Duration of unavailable or stale evaluation intervals against the declared policy and serving boundaries.
Unit	duration
Required inputs	policy-events; serving-observability

ASOR-SG-03: Decision-Time Policy Violation Rate

Decision: Measure governed workflows that breach declared controls.

Read-only sources: policy evaluation events; workflow and identity metadata; provider and model catalog.

Customer context: policy rules; severity mapping; evaluation window.

Availability: Unavailable policy evaluation is reported separately from zero violations.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-SECURITY-06.advisory_breach_count; METRIC-SECURITY-06.policy_evaluation_count; METRIC-ROI-43.decision_time_policy_breach_rate

Presentation contract: kind: breach_share_and_counts; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Count breached eligible evaluations once for the fraction. Raw breach findings may exceed the number of breached evaluations and retain their existing distinct output identity.

METRIC-SECURITY-06.advisory_breach_count

Property	Contract
Formula	Observed advisory breaches under the effective policy; this does not establish policy availability during an observation gap.
Unit	count
Required inputs	policy-events

METRIC-SECURITY-06.policy_evaluation_count

Property	Contract
Formula	Observed eligible advisory policy evaluations in the declared window.
Unit	count
Required inputs	policy-events

METRIC-ROI-43.decision_time_policy_breach_rate

Property	Contract
Formula	unique_eligible_decision_policy_evaluations_with_breach / eligible_decision_policy_evaluation_count
Unit	ratio
Required inputs	versioned_policy_evaluation_records; unique_eligible_decision_policy_evaluations_with_breach; eligible_decision_policy_evaluation_count; evaluation_coverage_and_effective_policy

ASOR-SG-04: Deployment Boundary and Zero-Egress Verification

Decision: Verify read-only, customer-resident, zero-payload and fail-open operating boundaries.

Read-only sources: deployment configuration; runtime egress evidence; retention evidence; fail-open health telemetry.

Customer context: deployment mode; allowed egress; retention contract; failure budget.

Availability: Configuration is not runtime proof; unchecked paths remain unknown and no raw content is required.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-SECURITY-03.processing_boundary_conformance_share; METRIC-SECURITY-03.processing_boundary_violation_count; METRIC-SECURITY-03.unchecked_processing_path_count; METRIC-SECURITY-04.retention_contract_conformance_share; METRIC-SECURITY-04.retention_finding_count; METRIC-SECURITY-04.unchecked_retention_surface_count

Presentation contract: kind: boundary_evidence_status; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: path_or_surface_id; path_type; declared_boundary; observed_destination_class; checked_at; evidence_reference; continuity_window; status; scope rule: Distinguish processing and egress path evidence from retention surface evidence. Approved external provider processing stays external; unchecked does not mean conforming.

METRIC-SECURITY-03.processing_boundary_conformance_share

Property	Contract
Formula	Checked eligible processing and storage paths conforming to the declared deployment boundary divided by eligible checked paths.
Unit	ratio
Required inputs	deployment-config; runtime-egress-evidence

METRIC-SECURITY-03.processing_boundary_violation_count

Property	Contract
Formula	Observed runtime boundary violations in the declared checked scope. External provider processing remains external.
Unit	count
Required inputs	deployment-config; runtime-egress-evidence

METRIC-SECURITY-03.unchecked_processing_path_count

Property	Contract
Formula	Count eligible paths marked unassessed in an inventory whose completeness has been established; missing runtime evidence does not establish compliance.
Unit	count
Required inputs	deployment-config; runtime-egress-evidence

METRIC-SECURITY-04.retention_contract_conformance_share

Property	Contract
Formula	Eligible checked stores, logs, traces and dumps meeting the declared metadata-only contract divided by eligible checked surfaces.
Unit	ratio
Required inputs	deployment-config; retention-evidence; integrity-tests

METRIC-SECURITY-04.retention_finding_count

Property	Contract
Formula	Observed retention-contract findings keyed to surface and evidence revision; canary results retain their bounded scope.
Unit	count
Required inputs	deployment-config; retention-evidence; integrity-tests

METRIC-SECURITY-04.unchecked_retention_surface_count

Property	Contract
Formula	Enumerated eligible surfaces explicitly untested in the coverage inventory; an unknown inventory cannot produce zero unchecked scope.
Unit	count
Required inputs	deployment-config; retention-evidence; integrity-tests

ASOR-SG-05: Regulatory Readiness Map

Decision: Map active workflows to declared regulatory and internal-control evidence.

Read-only sources: workflow registry; decision lineage; audit records; customer control inventory.

Customer context: jurisdiction; risk classification; control mapping; retention requirement.

Availability: Readiness is an evidence map, not a legal certification or claim of compliance.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-43.control_evidence_coverage

Presentation contract: kind: control_evidence_map_and_ratio; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: workflow_id; control_requirement_id; risk_tier; applicability_state; evidence_refs; evidence_validity; sufficiency_rule_version; evidence_state; scope rule: Package generation or validation counts cannot establish workflow-control evidence coverage or control effectiveness.

METRIC-ROI-43.control_evidence_coverage

Property	Contract
Formula	eligible_workflow_control_requirements_with_current_sufficient_evidence / eligible_workflow_control_requirement_count
Unit	ratio
Required inputs	versioned_workflow_control_requirement_inventory; risk_tier_mapping; evidence_sufficiency_and_freshness_rules; eligible_workflow_control_requirement_count; workflow_control_evidence_records_with_validity

ASOR-SG-06: Third-Party Model and Subprocessor Exposure Inventory

Decision: Know which providers, endpoints and jurisdictions receive enterprise AI traffic.

Read-only sources: provider and endpoint metadata; model catalog; network and egress metadata; subprocessor registry.

Customer context: approved provider list; jurisdiction policy; inventory period.

Availability: Inventory is bounded by observed read-only evidence; missing provider metadata is unknown.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-39.observed_third_party_entity_count

Presentation contract: kind: third_party_inventory_and_count; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: entity_type; entity_id; workflow_id; provider_endpoint_reference; subprocessor_or_jurisdiction_evidence; valid_at; source_revision; scope rule: Count each entity type separately. Provider and endpoint evidence remains independently usable without subprocessor or jurisdiction evidence; hostnames cannot establish either of those.

METRIC-ROI-39.observed_third_party_entity_count

Property	Contract
Formula	count_unique(observed_entities_by_declared_entity_type)
Unit	entity
Required inputs	typed_entity_identity_and_evidence; versioned_entity_identity_mapping; declared_entity_type_and_window

ASOR-C-01: Multi-Provider Traffic and Quota Utilization

Decision: Manage provider quotas, concentration and headroom.

Read-only sources: provider usage and quota metadata; OTel token metadata; account and region configuration.

Customer context: quota scope; provider account mapping; window.

Availability: Unknown quota state is shown as unknown; largest share and HHI are separate.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-CAPACITY-03.largest_turn_share; METRIC-CAPACITY-03.turn_hhi; METRIC-CAPACITY-03.largest_cost_share; METRIC-CAPACITY-03.cost_hhi; METRIC-ROI-36.observed_request_count; METRIC-ROI-36.observed_token_count; METRIC-ROI-36.request_quota_utilization; METRIC-ROI-36.token_quota_utilization

Presentation contract: kind: traffic_counts_utilization_concentration; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Totals remain available without quotas. Utilization requires a fresh positive quota with identical provider, account, model, region, quota kind, accounting basis and exact fixed or rolling window. Peak throughput and provisioned-capacity share are distinct statistics. Concentration retains its named turn or cost basis.

METRIC-CAPACITY-03.largest_turn_share

Property	Contract
Formula	Largest provider share of eligible observed turns.
Unit	ratio
Required inputs	usage

METRIC-CAPACITY-03.turn_hhi

Property	Contract
Formula	Sum of squared provider shares of eligible observed turns on the same population.
Unit	ratio
Required inputs	usage

METRIC-CAPACITY-03.largest_cost_share

Property	Contract
Formula	Largest provider share of eligible cost on the declared basis.
Unit	ratio
Required inputs	usage; prices

METRIC-CAPACITY-03.cost_hhi

Property	Contract
Formula	Sum of squared provider shares of eligible cost; missing rates do not withhold turn concentration.
Unit	ratio
Required inputs	usage; prices

METRIC-ROI-36.observed_request_count

Property	Contract
Formula	count_unique(eligible_request_occurrences)
Unit	request
Required inputs	tenant_qualified_request_occurrence_identity; request_receipt_metadata; declared_provider_account_model_region_window; source_population_coverage

METRIC-ROI-36.observed_token_count

Property	Contract
Formula	sum_unique(eligible_token_usage_by_request_and_token_kind)
Unit	token
Required inputs	tenant_qualified_request_usage_identity; observed_token_usage_and_kind; declared_token_accounting_rule; source_population_coverage

METRIC-ROI-36.request_quota_utilization

Property	Contract
Formula	eligible_quota_counted_requests_in_limit_window / effective_request_quota_for_same_limit_window
Unit	ratio
Required inputs	eligible_quota_counted_requests_in_limit_window; effective_request_quota_for_same_limit_window; provider_account_model_region_and_quota_kind; exact_fixed_or_rolling_limit_window; quota_effective_time_and_freshness; quota_accounting_and_population_coverage

METRIC-ROI-36.token_quota_utilization

Property	Contract
Formula	eligible_quota_counted_tokens_in_limit_window / effective_token_quota_for_same_limit_window
Unit	ratio
Required inputs	eligible_quota_counted_tokens_in_limit_window; effective_token_quota_for_same_limit_window; provider_account_model_region_and_quota_kind; exact_fixed_or_rolling_limit_window; quota_effective_time_and_freshness; token_quota_accounting_and_population_coverage

ASOR-C-02: End-to-End Workflow Latency Decomposition

Decision: Identify latency and wait-time drivers that affect accepted work.

Read-only sources: OTel span timing; routing and queue events; downstream write-back events.

Customer context: latency budget; workflow boundary; deadline.

Availability: Do not sum overlapping retrieval, orchestration, model or tool spans. Missing spans and incompatible clocks remain unknown; out-of-band ingestion lag is not request latency.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-24.workflow_elapsed_duration; METRIC-ROI-24.critical_path_stage_duration; METRIC-ROI-24.workflow_span_work_duration

Presentation contract: kind: workflow_duration_decomposition; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: workflow_run_id; stage_id; span_id; parent_span_id; dependency_edges; clock_basis; interval_start; interval_end; partition_policy; unclassified_interval; scope rule: Root elapsed duration remains available with incomplete traces. Disjoint critical-path stage totals reconcile with elapsed duration; parallel and nested span work remains separate. Request latency and serving-query duration are different grains.

METRIC-ROI-24.workflow_elapsed_duration

Property	Contract
Formula	workflow_terminal_at - workflow_started_at
Unit	second
Required inputs	matched_workflow_run_identity; workflow_started_at; workflow_terminal_at; clock_context

METRIC-ROI-24.critical_path_stage_duration

Property	Contract
Formula	sum(nonoverlapping_stage_intervals_on_declared_workflow_critical_path)
Unit	second
Required inputs	complete_workflow_dependency_dag; matched_stage_interval_metadata; versioned_critical_path_and_partition_rule; clock_context

METRIC-ROI-24.workflow_span_work_duration

Property	Contract
Formula	sum(duration_of_unique_spans_in_declared_span_population)
Unit	span_second
Required inputs	span_identity_and_parent_links; matched_span_start_and_end_times; declared_span_population_and_nesting_policy; clock_context

ASOR-C-03: Committed Cloud and Provisioned Throughput Burn-Down

Decision: Use or resize commitments while separating unavoidable and cancellable cost.

Read-only sources: cloud billing exports; commitment and PTU terms; usage and capacity telemetry.

Customer context: commitment term; coverage scope; cancellation boundary.

Availability: Savings claims require matched commitment and usage evidence; idle cost is not automatically avoidable.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-CAPACITY-07.committed_usage_coverage; METRIC-CAPACITY-07.committed_spend_coverage; METRIC-CAPACITY-07.realized_commitment_cost_delta; METRIC-CAPACITY-10.idle_committed_capacity_cost; METRIC-ROI-22.unavoidable_committed_cost; METRIC-ROI-22.cancellable_future_capacity_cost

Presentation contract: kind: commitment_components; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Separate observed commitment utilization, realized matched cost difference and idle committed cost. Idle cost is not automatically cancellable or avoidable.

METRIC-CAPACITY-07.committed_usage_coverage

Property	Contract
Formula	Compatible eligible billed usage quantity covered by valid commitments divided by eligible billed usage quantity. Missing invoice quantities withhold this output.
Unit	ratio
Required inputs	invoices; commitment-terms

METRIC-CAPACITY-07.committed_spend_coverage

Property	Contract
Formula	Eligible spend covered by valid commitments divided by eligible spend under the same amortization and term rules.
Unit	ratio
Required inputs	invoices; commitment-terms

METRIC-CAPACITY-07.realized_commitment_cost_delta

Property	Contract
Formula	Matched uncommitted contract-basis cost minus actual amortized committed cost, including unused commitment expense and explicit comparison assumptions.
Unit	USD
Required inputs	invoices; commitment-terms

METRIC-CAPACITY-10.idle_committed_capacity_cost

Property	Contract
Formula	Eligible amortized reserved/committed cost not consumed under explicit contract and utilization rules. Do not present provider-native parity functionality as a proven differentiator.
Unit	USD
Required inputs	invoices; commitment-terms

METRIC-ROI-22.unavoidable_committed_cost

Property	Contract
Formula	sum(non_cancellable_remaining_contractual_payment_obligations)
Unit	currency
Required inputs	contract_obligations; cancellation_terms; as_of_date; horizon

METRIC-ROI-22.cancellable_future_capacity_cost

Property	Contract
Formula	sum(cancellable_future_capacity_charges) - applicable_cancellation_penalties
Unit	currency
Required inputs	future_capacity_schedule; cancellation_terms; cancellation_penalties; as_of_date

ASOR-C-04: Fail-Open and Degraded-Mode Telemetry Continuity

Decision: Verify that attribution failures do not block or add material latency to production traffic.

Read-only sources: serving and interceptor telemetry; request outcome metadata; incident and deployment events.

Customer context: fail-open rule; latency budget; degradation window.

Availability: Continuity evidence is operational telemetry; it does not imply that a disabled collector was healthy.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-24.request_latency; METRIC-ROI-25.degraded_forwarding_continuity_rate; METRIC-ROI-25.degraded_request_latency_difference

Presentation contract: kind: degradation_continuity_and_latency; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: degradation_interval_id; collector_state; independent_timing_source; request_population_revision; fail_open_policy_version; evidence_coverage; scope rule: Scope request latency to independently observed degradation intervals. Supported duration survives an absent comparison baseline. Compare compatible means without claiming causation; a degraded collector cannot certify its own forwarding completeness.

METRIC-ROI-24.request_latency

Property	Contract
Formula	declared_statistic(request_end_timestamp - request_start_timestamp)
Unit	millisecond
Required inputs	request_boundary_events; request_latency_statistic; measurement_boundary

METRIC-ROI-25.degraded_forwarding_continuity_rate

Property	Contract
Formula	eligible_requests_with_evidenced_forwarding_during_degradation / eligible_requests_during_degradation
Unit	ratio
Required inputs	independent_request_receipt_and_forwarding_metadata; attribution_degradation_intervals; eligible_requests_during_degradation; versioned_fail_open_rule

METRIC-ROI-25.degraded_request_latency_difference

Property	Contract
Formula	mean_eligible_degraded_request_duration - mean_compatible_reference_request_duration
Unit	second
Required inputs	independent_request_timing_metadata; attribution_degradation_intervals; compatible_reference_request_timing; declared_workload_and_clock_comparison

ASOR-SU-01: Attributed Energy Consumption per Workflow

Decision: Estimate or report energy tied to accepted AI work.

Read-only sources: provider energy disclosures; customer energy telemetry; usage metadata; region and hardware factors.

Customer context: workload unit; factor boundary; region; period.

Availability: Measured and modeled energy are separate; incompatible or missing factors yield unavailable.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-44.measured_energy_total; METRIC-ROI-44.modeled_energy_total

Presentation contract: kind: energy_amounts; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Measured energy and factor-modeled energy stay separate under compatible allocation and physical boundaries.

METRIC-ROI-44.measured_energy_total

Property	Contract
Formula	sum(measured_energy_kwh_allocated_once_within_boundary)
Unit	kWh
Required inputs	energy_meter_or_qualified_measured_provider_records; physical_serving_boundary; allocation_policy; measurement_period

METRIC-ROI-44.modeled_energy_total

Property	Contract
Formula	sum(compatible_activity_quantity * qualified_energy_factor) + disjoint_modeled_boundary_energy
Unit	kWh
Required inputs	compatible_activity_quantities; qualified_energy_factors; factor_location_time_model_provenance; physical_boundary; allocation_and_uncertainty_model

ASOR-SU-02: Carbon Intensity per Accepted Result

Decision: Compare emissions intensity of workflows that produce accepted work.

Read-only sources: energy and carbon factors; provider or customer energy telemetry; workflow outcomes; region grid factors.

Customer context: accepted-result rule; carbon factor; region; maturity window.

Availability: No accepted outcome or compatible carbon factor means unavailable; human-time scenarios are not avoided emissions.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-ROI-45.measured_carbon_per_accepted_outcome; METRIC-ROI-45.modeled_carbon_per_accepted_outcome

Presentation contract: kind: carbon_per_accepted_result; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: ; scope rule: Measured and modeled carbon per mature accepted outcome remain separate. Carbon per dollar cannot substitute for this denominator.

METRIC-ROI-45.measured_carbon_per_accepted_outcome

Property	Contract
Formula	measured_carbon_total / unique_mature_accepted_outcome_count
Unit	kgCO2e/accepted_outcome
Required inputs	measured_carbon_total; unique_mature_accepted_outcome_count; cohort_acceptance_and_maturity_contract; matched_emissions_outcome_cohort

METRIC-ROI-45.modeled_carbon_per_accepted_outcome

Property	Contract
Formula	modeled_carbon_total / unique_mature_accepted_outcome_count
Unit	kgCO2e/accepted_outcome
Required inputs	modeled_carbon_total; unique_mature_accepted_outcome_count; cohort_acceptance_and_maturity_contract; matched_emissions_outcome_cohort

ASOR-SU-03: AI Energy Index Benchmark View

Decision: Compare workflow energy efficiency across models, routes and regions.

Read-only sources: provider environmental disclosures; customer energy telemetry; usage and outcome records; regional carbon factors.

Customer context: benchmark cohort; normalization rule; factor vintage.

Availability: Benchmark values are estimates unless measured energy is available; disclose factor provenance and uncertainty.

Freshness: required timestamps: source_event_at; observed_at; data_as_of; freshness limit: customer-configured source and workflow policy, with its version; states: fresh; stale; unknown; missing or stale behavior: Preserve supported observations with their original timestamps; withhold dependent current-period ratios when required evidence is stale or unknown.

Evidence states: states: observed; derived; estimated; unavailable; not_admitted; required result fields: metric_id; output_id; value; availability; reason; source_revision; definition_version; data_as_of; evidence_coverage; resolution: Resolve separately for every referenced output; a definition never supplies a result or inherits the strongest state from another output.; missing behavior: Return no numeric value and a specific reason for the affected output; preserve independently supported outputs.

Exact outputs: METRIC-SUSTAINABILITY-09.published_environmental_factor; METRIC-ROI-44.energy_benchmark_index; METRIC-ROI-45.carbon_benchmark_index; METRIC-ROI-44.measured_energy_per_accepted_outcome; METRIC-ROI-44.modeled_energy_per_accepted_outcome; METRIC-ROI-45.measured_carbon_per_accepted_outcome; METRIC-ROI-45.modeled_carbon_per_accepted_outcome

Presentation contract: kind: indexed_benchmark_and_trend; result rule: Every numerical result resolves independently from its exact qualified output contract. Missing prerequisite evidence withholds only that result.; record fields: benchmark_reference_id; reference_version; functional_work_unit; quality_boundary; system_boundary; factor_version; measurement_basis; period; qualified_underlying_intensity; scope rule: Retain underlying qualified intensity, a fixed compatible reference vintage and repeated-period trend. Separate measured and modeled bases and factor changes from operational changes; emissions per turn is not automatically this index.

METRIC-SUSTAINABILITY-09.published_environmental_factor

Property	Contract
Formula	Published comparable factors by model/hardware/region/time, with provenance, exclusions and uncertainty. Venturi hosts external evidence; missing or non-comparable factors do not justify a provider ranking.
Unit	gCO2e/token
Required inputs	energy-disclosures; carbon-factors

METRIC-ROI-44.energy_benchmark_index

Property	Contract
Formula	qualified_energy_per_work_unit / positive_compatible_reference_energy_per_work_unit
Unit	ratio
Required inputs	qualified_energy_per_work_unit; positive_compatible_reference_energy_per_work_unit; matched_workload_quality_and_system_boundary; factor_versions_and_uncertainty

METRIC-ROI-45.carbon_benchmark_index

Property	Contract
Formula	qualified_emissions_per_work_unit / positive_compatible_reference_emissions_per_work_unit
Unit	ratio
Required inputs	qualified_emissions_per_work_unit; positive_compatible_reference_emissions_per_work_unit; matched_workload_quality_and_system_boundary; factor_versions_and_uncertainty

METRIC-ROI-44.measured_energy_per_accepted_outcome

Property	Contract
Formula	measured_energy_total / unique_mature_accepted_outcome_count
Unit	kWh/accepted_outcome
Required inputs	measured_energy_total; unique_mature_accepted_outcome_count; cohort_acceptance_and_maturity_contract; matched_energy_outcome_cohort

METRIC-ROI-44.modeled_energy_per_accepted_outcome

Property	Contract
Formula	modeled_energy_total / unique_mature_accepted_outcome_count
Unit	kWh/accepted_outcome
Required inputs	modeled_energy_total; unique_mature_accepted_outcome_count; cohort_acceptance_and_maturity_contract; matched_energy_outcome_cohort

METRIC-ROI-45.measured_carbon_per_accepted_outcome

Property	Contract
Formula	measured_carbon_total / unique_mature_accepted_outcome_count
Unit	kgCO2e/accepted_outcome
Required inputs	measured_carbon_total; unique_mature_accepted_outcome_count; cohort_acceptance_and_maturity_contract; matched_emissions_outcome_cohort

METRIC-ROI-45.modeled_carbon_per_accepted_outcome

Property	Contract
Formula	modeled_carbon_total / unique_mature_accepted_outcome_count
Unit	kgCO2e/accepted_outcome
Required inputs	modeled_carbon_total; unique_mature_accepted_outcome_count; cohort_acceptance_and_maturity_contract; matched_emissions_outcome_cohort

Availability and customer feasibility

A view may combine customer telemetry, billing exports, system-of-record metadata, configuration, and independently maintained external factors. Each source carries freshness, scope, and evidence state. The customer-resident connector boundary is read only, and the view remains usable at the strongest supported grain when one layer is missing. Missing layers are labeled and withheld from dependent ratios rather than silently imputed.

The customer supplies the tenant scope, workflow definition, and acceptance rule for the relevant decision. Customer systems can usually provide the measurable inputs through logs, traces, provider usage and billing exports, identity and repository records, workflow or ticket status, and accepted outcome records. External factors are permitted only where the view declares them, such as grid intensity or provider price metadata, and remain separately labeled.

Economic anchors

• Cost per Accepted Result: eligible paid workflow cost divided by unique customer-accepted business outcomes. Gross cost per invocation remains a separate diagnostic.
• Net Realizable Benefit: attributable benefit less included cost, with cash savings and released capacity reported separately. Released capacity is not presented as payroll savings without customer evidence.
• Monthly Attribution Reconstruction Effort: recurring human time spent reconstructing ownership, cost, and outcome relationships before and after the record is available. Time saved is an operational measure, not an automatic financial ROI claim.

Source binding

The platform harness task ASOR-VIEWS-01 binds this manifest and these documentation requirements. App and demo surfaces expose the same decision views and existing metric identities. Demonstration values remain labeled as illustrative; customer measurements retain their source and evidence state. Investor-facing language preserves the present-tense category, domain selection, and economic anchors.

Give feedback on this page

Sign in to send feedback. Opens in a new tab.

Copy page details

Docs feedback: AI System of Record and the 42-view contract
Page: https://docs.venturi.systems/engine/ai-system-of-record/