Skip to content

The Roastery Lab · Evidence

The proof, labeled

Transparency you can trust because every artifact says what it is.

Publishing is a governed projection of approved work — not a mirror of the private database. Three kinds of artifact live here, and they are never conflated: a reference profile is a target, a batch report is a measurement, an experiment report is a finding.

Reference profile

Target

The approved target a coffee is roasted toward — its intended cup, phases, and events. A design intent, clearly labeled as a target, not a record of one batch.

Batch evidence report

Measured

An actual roast: its date, lot, capture tier, measured anchors and curve, weight loss, and QC result. Published only after it happens and is approved.

Experiment report

Published finding

A controlled study: hypothesis, control, one variable, linked batches, sensory results, limitations, and the production decision — including an inconclusive one.

Reference profilesApproved target

The seven roasts, each with its approved target profile — the standard every future batch will be measured against.

Batch evidence reportsNone yet

No production batches have been roasted yet — Pacific Roots is pre-launch, and we won't publish a batch report until a real roast happens and passes its QC gate. When the first batches run, each one's measured anchors, curve, weight loss, cupping result, and capture tier will publish here — dated, versioned, and superseded rather than silently overwritten.

The honest version of “every coffee ships with its evidence” — coming, not claimed.

Experiment reportsIn progress

E-001 · The rest-window study

The Lab's first planned original dataset: does the same curve peak at different rest days by roast level? When it runs, its full report — hypothesis, protocol, linked cuppings, and conclusion — publishes here.

See the research →

Browse the coffees →