The SubstanceIQ Methodology

Substance, measured. SubstanceIQ is a formal measurement and classification framework for economic substance under the OECD DEMPE rules — the first instrument built for the question every multinational must now answer annually: which entities actually perform and control the functions that create value, and can you prove it?

Two numbers per entity

Every group entity receives a Substance Index (0–100) — what it actually contributes, scored across twenty industry-tuned DEMPE activities — and a Confidence Index (0–100) — how much of that contribution is supported by evidence. The gap between the two is your audit exposure, quantified question by question.

Four dimensions, not one impression

Each activity is scored on four behaviourally anchored dimensions: Control (who decides, with what finality), Capability (who has the people and systems), Execution (who does the work), and Risk (who bears the consequences in conduct, not just on paper). Scale levels are defined by observable conduct — approval records, repository logs, loss bookings — not adjectives.

Classification with gates

The Substance Index places each entity in one of four tiers, from routine provider (T1) to entitled to residual returns (T4). Admission to the higher tiers is non-compensatory: no volume of activity reaches ownership without demonstrated control and risk — the exact logic of the OECD Guidelines, enforced arithmetically. Gates demote; they never promote.

Machine-enforced coherence

Twelve coherence rules run across the whole group: one final decision-maker per activity; claimed work cannot exceed the work that exists; risk requires financial shoulders; authority requires people able to exercise it. Impossible stories cannot be filed — by anyone in your organisation. The rules police impossible groups, never misaligned ones: a genuine misalignment is a finding to report, not an error to hide.

Evidence, graded

Every claim carries an evidence grade under four criteria — contemporaneity, source independence, attribution, verifiability. The strongest single document determines the grade; weak evidence does not stack; yesterday's memo about last year grades zero. Audit-ready status is gated on confidence: large claims demand strong proof.

Deterministic by design

The assessment engine contains no artificial intelligence. Identical inputs produce identical outputs, including every narrative sentence — because anything you defend before a tax administration must be reproducible. AI exists only in optional, separately switchable add-ons that assist and never decide, operating on anonymised data through EU-based services.

Regime interfaces

The framework speaks directly to the bright-line regimes: it produces the Amount B scoping evidence the OECD's own pricing tool states it cannot test; it screens and monitors hard-to-value intangibles year by year, building the exemption defence that cannot be assembled retrospectively; and it documents what Pillar Two's mechanical carve-out does not measure.

Versioned, governed, published

The methodology's architecture is published to peer-review standard (methods article prepared for the World Tax Journal, IBFD; companion architecture at the International Transfer Pricing Journal). Parameters are versioned data: every assessment permanently records the exact parameter set that computed it, locked years reproduce identically forever, and methodology upgrades never silently recompute your history. The full question bank, industry weight configurations and calibration data are proprietary and protected as trade secrets — the architecture is public; the instrument is ours.