Evidence before adjectives.
The same report identifies the same diamond. Everything else needs evidence. Here is what this foundation can establish, and what still needs to be validated.
One grading lab. One report number.
We match a stone by the combination of its grading lab and normalized report number. We preserve lab-grown prefixes. Matching a shape, carat, color and clarity alone is not enough to say two listings represent the same stone. Bare report numbers do not prove a grading lab.
The lowest complete total comes first.
The ranking engine adds the listing price, shipping, tax and duty when each is known for the buyer’s destination. A listing with missing costs cannot win a lowest-landed-price claim. Commission rates are not an input to the price order. Every number currently shown on this site is illustrative, including the assumed costs.
Some listings should never become a deal.
Our initial gates reject stale observations older than 72 hours, invalid timestamps, incomplete totals, contradictory origin identifiers, and prices below half of an applicable cohort median. A cheap price is not proof of lab-grown origin, and the absence of an LG prefix is not proof of natural origin. Licensed feed evidence is required for both. Held listings are excluded from comparison ladders.
Compare within the right cohort.
Magic-size carat boundaries keep 0.99 ct separate from 1.00 ct, and 1.49 ct separate from 1.50 ct. Benchmarks will also separate origin, grading lab, shape and grades. GIA lab-grown tier reports have their own grading-system field. We are not publishing a median or hedonic estimate from this tiny sample catalog.
An angle is not a light-return score.
Report proportions can describe geometry. They cannot justify a precise optical score without a calibrated model, uncertainty bands and independent validation. Fancy-shape bow-tie and eye-clean assessments need visual evidence. Until calibration and evidence are available, these scores remain unavailable. The interactive diamond is artwork, not a measured optical result.
Optical twins need more than similar grades.
The planned twin search compares origin, shape, measured length and width within a narrow range, then applies a calibrated cut floor and evidence checks. It does not need an embedding database. No current sample is claimed to be visually indistinguishable from another diamond.
Independent review, when it is ready.
The planned concierge service is a separately priced review for larger natural-diamond purchases. Its scope includes report and inscription checks, optical evidence, seller terms and complete prices. It is not bookable yet, no reviewer has been represented as engaged, and no fees are being collected.
What has to happen before deal badges appear.
The launch plan requires at least 50,000 deduplicated stones, a manual spot-check of 100 random stones, calibrated optics, and an optical score band meeting the minimum cut threshold. This foundation has none of those live-data prerequisites. It deliberately publishes no deal badges or claimed fair-value scores.