15
Method / Disclosure

How this atlas
is actually made.

The editorial method behind The Chocolate Atlas: how claims are sourced and scoped, how AI assistance is used and bounded, what is reviewed and when, and how to challenge anything published here.

A field guide is only worth as much as its method: what it checked, what it refused to claim, and how it would know if it were wrong.

Why this page exists

Readers deserve to know how anything they are asked to trust was produced — including the parts that are uncomfortable to state plainly. Everything below is a description of the current process, not an aspiration.

The process

Five commitments

These govern every route on the site, including the ones that would read more confidently without them.

01

Every claim is scoped before it is written

A claim is only written when it can be stated at a scope a source actually supports. “Fermentation changes seed chemistry” and “this origin tastes of red fruit” need different kinds of evidence, and the second is written as a prompt rather than a promise. Where the evidence is thin, the sentence says so instead of being smoothed over.

02

Sources are chosen for what they can prove

The 44 records on the research desk favour standards bodies, international and national public institutions, peer-reviewed research and botanical references. First-party company information is used only for narrow company facts, never for company impact claims. Each record carries its publisher, evidence kind, the scope of what it can support, and the date it was last reviewed.

03

AI assistance is used for drafting, never for evidence

Research, drafting and editing on this site are AI-assisted. Every draft is written against a collected source packet, then checked line by line against those sources by a human editor before publication. AI is not used to generate citations, to decide what a source says, or to fill a gap where evidence is missing — a gap is left visible instead. Where a number, a standard or a regulation is quoted, it was read in the original document.

04

Imagery is interpretive and labelled as such

The editorial photography and diagrams on this site are generated or hand-drawn. They illustrate a mechanism or a situation; they are never documentary evidence of a named person, farm, batch or historical event, and captions say so. The twenty process diagrams are drawn from the process descriptions they sit beside, not from a photograph of a specific facility.

05

Corrections are welcome and are treated as evidence

If a claim here is wrong, out of date, or stated at a wider scope than its source supports, that is a defect. Write to the address below with the page and the source you would use instead. Corrections change the visible copy and the source record together, so the two can never drift apart.

01

No first-hand tasting is claimed

Nothing on this site reports a bar the authors personally tasted, a farm they visited, or a batch they measured. Sensory material teaches a method and a vocabulary; it does not review products.

02

No ranking, no sponsorship

No brand, retailer or certifier has paid for placement, and no affiliate revenue is taken. Inclusion in the brand index is educational and is not an endorsement of sourcing practice.

03

Not advice

The quality and safety material summarises published guidance for general education. It is not medical, dietary, legal or regulatory advice, and it does not replace a manufacturer's current statement about a specific product.

04

Evidence ages

Sustainability programmes, ownership, regulation and market data change. A review date on a record is a statement about when it was last checked, not a guarantee that it is still current today.

Challenge this

Found a claim that
outruns its evidence?

Send the page, the sentence and the source you would use instead. Corrections are made in public and update the source record at the same time.

contact@thechocolateatlas.com