How We Match Diamonds When Certificate Numbers Are Missing

Certificate numbers make diamond matching trivial. String comparison, deduplication, done in milliseconds. But across the 9.8 million unique stones in our live index, a meaningful fraction of listings arrive without a usable cert number. Mistyped. Truncated. Deliberately withheld. Sometimes the stone simply isn't certified at all.

Our cross-retailer fingerprinting system covers the full methodology for identifying the same physical diamond listed at different prices across multiple retailers. If you want the broad picture of how we deduplicate the market, start there. This post focuses on the harder technical problem: what happens when the certificate number, the single easiest identifier, isn't available.

Three ways cert numbers fail

The first scenario is deliberate suppression. Retailers strip or obscure cert numbers from their listings because they don't want you searching that number on a competitor's site and finding the same stone for less. We see this regularly across midmarket online retailers who source from identical wholesalers as everyone else but prefer you don't notice.

Second: data entry errors. A GIA report number is typically ten digits. One transposed digit and the match fails silently. We've observed truncated numbers, numbers with appended retailer suffixes, and numbers reformatted with spaces that break exact string matching. The stone is certified. The number is technically present. It just doesn't match anything.

Third: uncertified stones. Plenty of diamonds trade without GIA, AGS, or IGI reports. Smaller stones under 0.30ct, older estate pieces, and some retailer branded lines ship with in house grading only. No independent cert number exists for these stones. A matching system that relies solely on cert numbers treats them as unmatchable. We don't.

Thirteen fields, zero ambiguity

When cert number matching fails or isn't possible, we construct a gemological fingerprint from the stone's physical and grading characteristics. Thirteen fields, each contributing to a composite identity:

Field	Precision	Role in matching
Shape	Exact category	Primary filter, eliminates 85%+ of candidates
Carat weight	To 0.01ct	Narrows candidate pool dramatically
Colour grade	Single grade	Core grading identity
Clarity grade	Single grade	Core grading identity
Cut grade	Single grade	Round brilliant differentiator
Polish grade	Single grade	Secondary differentiator
Symmetry grade	Single grade	Secondary differentiator
Fluorescence	Intensity level	Eliminates near misses
Length (mm)	To 0.01mm	Physical measurement lock
Width (mm)	To 0.01mm	Physical measurement lock
Depth (mm)	To 0.01mm	Physical measurement lock
Table %	To 0.1%	Proportions verification
Depth %	To 0.1%	Proportions verification
Grading lab	Exact match	Prevents cross-lab confusion

The physical measurements are the backbone of the whole system. Two round brilliants can share identical grades across every single category (colour, clarity, cut, polish, symmetry, fluorescence) and still differ at the measurement level. A stone measuring 6.52 x 6.55 x 4.03mm is not the same diamond as one at 6.51 x 6.54 x 4.02mm, even if everything else lines up perfectly.

Measurements function as the closest thing to a physical fingerprint short of plotting individual inclusions. Manufacturing variance means even two diamonds cut to identical target proportions will diverge at the hundredth of a millimetre. That divergence is our signal.

We require all thirteen fields to match simultaneously when no cert number is present. Not twelve. Not "most of them with a fuzzy tolerance." All thirteen. This is conservative by design.

When two diamonds look identical on paper

The obvious risk with fingerprint matching is false positives: declaring two different physical stones are the same diamond because they happen to share identical specifications. At the scale of our index, this isn't theoretical. Popular specification combinations produce many similar stones.

Consider the scenario. A 1.00ct G VS2 round brilliant, Excellent cut, Very Good polish and symmetry, None fluorescence, graded by GIA. There are hundreds of stones in the active market matching those seven criteria. Without the physical measurements, fingerprint matching would collapse into chaos at these popular nodes.

Our guards work in layers.

Measurement precision eliminates most collisions immediately. Two stones must agree on length, width, and depth to the hundredth of a millimetre. Even at that popular G VS2 specification, the measurement combinations spread stones out into distinct identities.

Table and depth percentage act as a secondary consistency check. These values are derived from the physical measurements but reported independently on certificates, so they serve as a verification layer. If measurements match but table percentage doesn't, something is wrong with the data.

When we do encounter multiple stones with identical fingerprints across all thirteen fields from separate supply chains (not the same stone at multiple retailers, but genuinely different diamonds), we flag these as provisional rather than confirmed matches. Our system distinguishes between high confidence identifications and cases requiring additional verification. In practice, full thirteen field collisions between genuinely different stones are exceptionally rare. The measurement precision at 0.01mm makes it so.

Half the market sits on multiple shelves

The cross-retailer duplication in the diamond market is enormous. Most buyers don't realise they're looking at the same stones everywhere.

Metric	Count	Share of index
Unique diamonds in index	9,812,685	100%
Listed at 2+ retailers	5,313,632	54.2%
Listed at 5+ retailers	1,560,887	15.9%
Listed at 10+ retailers	150,458	1.5%

More than half the diamonds in our index appear on multiple websites. Over 150,000 stones sit on ten or more retail sites simultaneously. Same physical diamonds, sourced from the same wholesalers, often photographed with identical imagery, listed at wildly different prices.

The stones appearing at ten or more retailers cluster heavily in the engagement ring sweet spot: 1.00 to 2.00ct, D to H colour, VS2 to SI1 clarity, round and oval shapes. These move through large wholesale networks and every participating retailer lists the full feed. A cert only matching system finds most of these duplicates without difficulty. But when even one retailer in the chain strips the cert number, that listing becomes invisible to cert only comparison tools.

Our fingerprint matching captures these orphaned listings and reconnects them to their duplicate group. For buyers, this matters specifically because the retailer who stripped the cert number is frequently the one offering the lowest price. They compete aggressively on cost. They just don't want you confirming the identity independently. The advanced search on CaratHunter surfaces these cross-retailer matches regardless of whether the cert number was provided.

No cert, no measurements, no confidence

Uncertified diamonds present a fundamentally different challenge. Without a lab report, you're missing guaranteed accuracy on colour and clarity grades, standardised measurement reporting, and the cert number itself. Our certification score system assigns a confidence rating to each listing based on how complete and verifiable its data is.

For fingerprint matching to work on uncertified stones, the listing needs to provide physical measurements to hundredth of a millimetre precision. Many don't. When a retailer lists a stone as "1.02ct G VS2 Round" with no measurements, no cert, and no proportions data, we can't fingerprint it. It sits in our index as an isolated listing, unable to be compared against anything.

This isolation is itself useful information. If a stone can't be fingerprinted, that's a signal about the listing quality. Nobody can verify whether the price is competitive because nobody can confirm what the stone actually is at the physical level. Stones with full measurement data and no cert remain matchable and comparable. Stones with neither are essentially unverifiable through any automated system.

We treat the presence of physical measurements, independent of certification, as a minimum threshold for meaningful online diamond comparison. A retailer providing dimensions of 6.52 x 6.55 x 4.03mm tells you something concrete and checkable. A retailer providing only "approximately 6.5mm" tells you almost nothing useful for identification purposes.

The buyers who benefit most

Diamond identification without cert numbers matters most for two groups. Buyers shopping uncertified stones (common in smaller carat weights and estate jewellery) who need some basis for price comparison. And buyers who've found a great price at a retailer that strips cert data, who want to verify the stone's identity through its physical characteristics before purchasing.

For the first group, fingerprint diamond matching provides the only cross-retailer comparison pathway available. Without it, each uncertified listing exists in a vacuum. With it, identical stones surface together and price outliers become visible.

For the second group, the fingerprint serves as confirmation. You find a 1.50ct F VS2 oval at a surprisingly low price, but the listing shows no cert number. The measurements read 8.41 x 5.92 x 3.67mm. If that exact measurement combination appears at three other retailers with a GIA number attached, you've got your identification. The uncertified diamond comparison becomes a certified one through triangulation.

What we're watching next

The fraction of retailers suppressing cert numbers has grown over the past twelve months. Not dramatically, but the trend is consistent. As more comparison tools enter the market, more retailers respond by removing the easiest matching identifier. Our thirteen field fingerprint was built for exactly this arms race. The measurements don't lie, and retailers can't hide them without making their listings unsellable. A diamond without dimensions is a diamond nobody buys online.

We're also tracking measurement reporting precision across the index. Some retailers round to 0.1mm rather than 0.01mm, which degrades fingerprint confidence. Listings with rounded measurements get a lower match confidence score and a lower certification score overall. The market is slowly standardising on higher precision reporting, which makes our job easier with every quarter.