Methodology Statement

1. Purpose and principle

Our guiding principle is simple: we present evidence, not a verdict. We surface public records, apply a transparent and published method, and show every step. We do not certify a product, approve it, or judge its sustainability, authenticity, or legal compliance. We give a reader the underlying record and the working behind a number. The reader, the seller, and the buyer draw their own conclusions.

Three commitments follow from that principle, and they govern everything below:

1. We are a conduit to public records. The provenance and trade figures we surface originate in public government ledgers. We structure and present them. We are not their author, and we do not warrant their underlying accuracy.
2. The seller owns the claim. A seller decides what to represent to its customers and is responsible for that representation. We supply the record and its working, not permission to make a claim.
3. We always show the working. Every figure is versioned, timestamped, and traceable to its source. That traceability is the product.

2. What the $MMR is, and how it is computed

The Material-to-Marketing Ratio ($MMR), also called Under the Price, compares an object's estimated material-and-labour cost against its retail price. Its purpose is to make visible the gap that marketing usually conceals: the distance between what an object plausibly costs to make and what it is sold for.

The $MMR is not an audit of a company's books. It is an estimate built from public commodity-trade data and a set of stated assumptions. The material and labour figures are estimates derived from public commodity data and openly stated assumptions, not audited company financials. No brand's internal cost sheet is used, requested, or implied.

In plain language, the calculation works in three steps:

1. Identify the materials. For a given object we identify its principal raw materials (for example, a specific leather, silk, or a regulated exotic material) and the approximate quantity each contributes.
2. Estimate a material-and-labour cost. For each material we attach a per-unit cost estimate drawn from public commodity-trade records, then add a labour estimate based on stated assumptions about the work involved. The sum is the estimated material-and-labour cost.
3. Compare against retail price. We express the relationship between that estimated cost and the object's stated retail price as the $MMR.

The formula, stated for transparency and marked as provisional:

A worked, illustrative example appears in Section 7. Every input marked as an estimate above is exactly that. Where a figure is illustrative rather than computed from live data, we mark it as illustrative on the surface where it appears.

3. Data sources

Two public government data sources underpin the provenance and cost layers. Both are public records, both are provided to us and to everyone else "as is," and both can change after we read them.

CITES (Convention on International Trade in Endangered Species of Wild Fauna and Flora)

CITES publishes trade records for regulated and exotic materials, for example species-controlled skins and other protected materials. We use it to associate a regulated raw material with the public record of its recorded trade. CITES data reflects what was reported to the Convention and is updated on the Convention's own schedule, not ours.

UN Comtrade

UN Comtrade is the United Nations database of international commodity trade, organised by Harmonised System (HS) codes. We use it for commodity-level cost reference, for example HS chapters 41 and 42 (leather and leather goods) and HS chapter 50 (silk). Comtrade reflects trade reported by member countries and is revised periodically as countries submit and correct their filings.

The OAC corpus and methodology

Alongside the two public ledgers, we apply the OAC research corpus and this published methodology to structure the raw data into a comparable figure. This is our contribution: the structuring and the stated assumptions. It is documented here so it can be examined.

On currency and change. These are living government datasets. A figure we publish is accurate as of the timestamp attached to it and no later. Underlying records may be revised, restated, or corrected by their publishers after we read them. We do not warrant the accuracy, completeness, or currency of the underlying public data. We warrant only that we have surfaced it faithfully and shown its date.

4. Provenance tracing

Provenance here means raw-material origin traced to public trade records. It is not a full chain-of-custody certification, and we do not present it as one.

For a given object and its identified principal materials, we locate the corresponding entries in the CITES and UN Comtrade records and associate the object's raw materials with the origin and trade information those public records contain. The result is a statement of the form "this class of raw material is recorded in these public trade records," with the records and their dates shown.

What this method can and cannot do:

• It can associate a raw material with the public trade record that pertains to it, and show that record and its date.
• It cannot prove that one specific physical object contains material from one specific shipment. It is a link between an object's material class and the public record, not a certified chain of custody for the individual item.

We state this limit plainly wherever provenance appears, so the association is never mistaken for a custody guarantee.

5. Limitations and assumptions

This section is the heart of the document. Read it before relying on any figure.

• Estimates, not certified figures. The material and labour costs, and therefore the $MMR, are estimates built from public commodity data and openly stated assumptions. They are not audited company financials, not certified valuations, and not a brand's actual costs. Real costs vary with contracts, volume, quality grade, and factors no public dataset captures.
• Illustrative where marked. Some figures and examples in our published materials, including the worked example in Section 7, are illustrative rather than computed from live data. These are labelled illustrative on the surface where they appear. An illustrative figure demonstrates the method; it is not a measurement of a real object.
• Sources are accurate only as of their timestamp. Every figure is tied to the date it was drawn. Public sources change after that date. A figure is a snapshot, not a standing guarantee.
• No claim about a product's real-world properties. An $MMR figure describes a relationship between an estimated cost and a retail price. It is not a statement about a product's actual sustainability, environmental impact, ethics, authenticity, quality, or legal compliance. A high or low ratio implies none of those things.
• Not a certification. OAC Colophon is not a certification body, a conformity-assessment body, or an accredited verifier. Nothing here certifies, approves, endorses, or guarantees any product.
• Not legal or regulatory advice. This methodology and the figures it produces are information, not legal, regulatory, or professional advice, and are not a substitute for it. They do not determine or guarantee any party's compliance with any law or regime.
• Assumptions are stated, not hidden. Where the method depends on an assumption (for example, a labour estimate, a quantity estimate, or a mapping from an object to an HS code), we state the assumption rather than bury it. Different reasonable assumptions can produce different figures. Showing them is the point.

6. Versioning and timestamps

Defensibility depends on knowing exactly which method produced which number, and when.

• Every figure is dated. No figure ships without a timestamp for the underlying data.
• Every figure is tied to a methodology version. This document carries a version number. A figure produced under a given version can be traced back to the exact method described in that version.
• Corrections are handled openly. When we find an error in our method, or a public source revises data we relied on, we update the figure, re-date it, and note the change. We do not silently overwrite a published figure without a record that it changed.
• Superseded figures remain traceable. A reader who saw an earlier figure should be able to see that it was produced under an earlier method version and has since been updated.

7. How to read a figure (illustrative)

The following example is illustrative. The numbers are placeholders chosen to show the method, not measurements of a real object.

How to read that responsibly:

• The ten-times figure is a relationship between an estimate and a price, not a claim that the object is overpriced, unethical, or poorly made. A large gap between cost and price can reflect brand, design, retail margin, and many legitimate factors. The figure exposes the gap; it does not judge it.
• Every input above is an estimate or an illustration, labelled as such.
• The real figure, when computed, would carry its source (CITES, UN Comtrade, OAC methodology), its timestamp, and its methodology version, exactly as required before any figure ships.

8. The refusal rules

An instrument is only as good as what it declines to say. These are the conditions under which the Collection returns nothing rather than something, stated so that a reviewer can check them against the behaviour.

1. An answer that cannot be grounded is withheld, not generated. Retrieved passages are scored against the question, and if the best score falls below a calibrated floor the reading is refused. The floor is 0.47 cosine similarity (app/config.py, arena_min_score), calibrated against all-MiniLM-L6-v2 where on-topic questions score 0.55 to 0.74 and off-topic 0.21 to 0.38. The refusal is a branch in the composer, not a prompt instruction (app/rag.py, compose()), so it cannot be talked around.
2. The refusal says so in plain words. The customer receives: "I cannot ground an answer to this question in the corpus. Retrieval did not surface sufficiently relevant sources, so I am withholding rather than inventing a claim." No partial answer, no hedged paraphrase, no citation list assembled from passages that did not qualify.
3. A withheld question is not billed as a grounded one. Only a grounded answer consumes a metered question.
4. An object with no independently documented cost basis carries no figure. It is published with its exclusion and the reason, rather than with an estimate presented as a measurement.

9. Negative results, including one that disproved our own assumption

Eleven objects are published. Three carry a $MMR. The other eight are published without one, each with the reason it was withheld. That ratio is the method working, not the dataset being thin: a figure requires an independently documented retail price and an independently sourced cost basis, and most objects do not have both.

• Exotic crocodile tote — 4.2×
• Silk twill scarf, 90cm — 4.5×
• Luxury eau de parfum — 3.6×

The gold cuff is the one we would most like to have omitted, and it is the reason to trust the other three. An 18k solid-gold cuff was expected to show the same pattern as the rest. Anchored to roughly 45g of 18k gold at a documented spot price and to a published retail band, the material content approached the retail price: a multiple near 1×. There was no markup of the kind the method exists to expose.

We published it. The assumption that a luxury object conceals a large material-to-retail gap is false for gold, where the material is itself the store of value and its price is public and unarguable. An instrument that only reports findings which flatter its own thesis is not an instrument. This is the negative result, and it is kept in the dataset rather than in a footnote.

10. What we deliberately do not claim

For the avoidance of doubt, and consistent with the OAC Colophon Claims and Wording Standard, OAC Colophon does not make, and does not enable a customer to make, any of the following claims on the basis of our figures:

• No environmental-outcome claims. We do not state or imply that any product is sustainable, eco-friendly, green, conscious, responsibly crafted, environmentally friendly, climate neutral, carbon neutral, net zero, or planet positive. The $MMR says nothing about environmental impact.
• No certification or endorsement. We do not certify, approve, or endorse any product, and we are not a seal of approval. Colophon is not a certification or conformity-assessment body.
• No compliance guarantee. We do not make any party compliant, guarantee compliance, or assure regulatory approval under the EU Empowering Consumers Directive, the Ecodesign for Sustainable Products Regulation or Digital Product Passport, CSDDD, EUDR, or any other regime. Our figures are evidence a party may use in support of its own claims, and nothing more.
• No warranty of underlying truth. We do not guarantee that a product is authentic, that our estimates match a brand's real costs, or that public source data is complete or current.

What we do claim is narrow and, we believe, defensible: we have surfaced identifiable public records, applied a stated method, shown the working, and dated the result. The conclusion is the reader's to draw.