How to Buy Gemstones: The Six Checks That Decide the Price

Part of Gem Lab Reports and Treatments

Most pages that answer "how to buy gemstones" are shops selling stones, plus a forum thread that recommends other shops. None of them says what to check before the money leaves. This one names no seller, recommends no stone, and works only from documents.

There are six checks, in the order they move a price. Each one is a document to demand or a question to ask, and each can be verified without trusting the person who quotes it. We do not give investment advice.

Learning how to buy a gemstone is mostly learning which documents to demand. Everything below is a document, or arithmetic on one.

The first control makes the others possible: an independent lab has to be named, and its report has to be checkable.

Each lab runs its own lookup, and each asks for different details.

A lookup that returns a record proves less than buyers assume. GIA says Report Check confirms that the information on a report matches what is archived in its database. The labs state the limit themselves. Gübelin calls its online data informational and non-binding, and says its database covers documents under 10 years old. GRS says finding report data online does not guarantee the authenticity of the report in hand, and that authentication needs the original document plus a re-inspection of the stone. SSEF says a report is valid only with its original signatures, stamps and label.

So run three comparisons, not one: paper against database, paper against stone (weight, measurements, shape, color), and the report's date against today. A report describes the stone on the day it was examined. SSEF notes that an emerald's filler grade records the condition at the time of testing, and in June 2026 it warned, as National Jeweler reported, that some emeralds are cleaned, tested and refilled after a report is issued.

One more limit matters for price. No lab values a stone. A report names the species, states the treatment and, sometimes, the origin. Nothing on it is a price. How to verify a gem lab report covers the lookup lab by lab.

The treatment line decides the bracket, so read the wording on the report rather than the summary a seller gives.

GIA draws a line that decides which object is being sold. Glass filling fractures in a coherent natural ruby is still described as natural ruby with clarity enhancement. Glass that binds ruby fragments produces a composite, and the report says so. A composite is not an expensive ruby; it is a different material. See glass-filled ruby.

There is also a legal floor. Under the FTC's Jewelry Guides, 16 CFR 23.24, a treatment must be disclosed where it is not permanent, where it creates special care requirements, or where it significantly affects value. Where the buyer cannot examine the stone before purchase (online, from a catalog, on a screen), the disclosure belongs in the solicitation or product description. A listing with no treatment line is short of that standard whatever the stone turns out to be.

This is where buying an emerald or a ruby starts: the filler line and the heating line, read before the color is discussed. For what each wording does to a price, see the guide to gem lab reports and treatments, no indications of heating and minor oil in emerald; for the species itself, what emerald is and what ruby is.

Origin is the line buyers pay most for and check least.

Gübelin describes geographic origin as a professional opinion about a stone's probable origin, formed by comparing its features with reference stones. Its terms say the opinion is not an opinion on quality or value, carries no warranty on the actual origin, and is left out where the evidence is insufficient. That is the strongest form this statement takes in the trade, and it is still an opinion.

So the check is narrow: which lab, which report number, and what exactly does the origin line say. A dealer's invoice, an online listing and an auction catalog note are all claims by an interested party. Until a named lab carries the line, an origin is worth zero in the price, however convincing the story attached to it.

Origin does separate stones inside a band. Two stones of the same size, color and treatment from different sources are not the same asset, and buyers pay for the difference. The comparison is only valid when both stones carry a lab line. The origin sections of what sapphire is set out that hierarchy, and SSEF vs. Gübelin vs. GRS vs. GIA compares how the labs word it.

That is also the whole of how to buy sapphires at the fine end: an origin line from a named lab, on a stone whose treatment line has already been read.

A per-carat price is not a price until size, treatment and lab are fixed.

Price per carat rises with size, in steps at roughly 1, 2, 3 and 5 carats. A 2.9-carat and a 3.1-carat stone of the same quality can sit in different brackets. Above 5 carats, fine stones are priced one by one, usually at auction. Compare a quote with a stone two carats away and the comparison means nothing.

The benchmark has to come from a named sale, not from a seller's own history. Colored gemstone prices per carat gathers Carat Brief's per-carat pages, each built on auction results and producer disclosures: ruby, sapphire, emerald, tanzanite, alexandrite and Paraíba tourmaline, the last two on thinner auction evidence. Spinel, garnet and other tourmalines have species guides (spinel, garnet, tourmaline) but no per-carat page yet.

Two traps sit inside the benchmark. A record is a ceiling, not a target: the highest per-carat result for a species names one exceptional stone at one public sale, and describes nothing about a two-carat stone with a moderate oil line. And a ring's price is not a stone's price: divide the total by the carat weight and the mount, the maker and the diamonds are inside the per-carat number.

This is why buying loose gemstones is the easier problem. With no mount in the figure, the ask can be tested directly against a band, and a stone that looks fine at a commercial per-carat price has to be explained.

The displayed price is not the price paid. Each channel adds its own layer on top.

At auction, Christie's and Sotheby's publish the same ladder, effective in 2026: 28% on the first $2,000,000 of hammer price, 22% on the portion from $2,000,000 to $8,000,000, and 15% above that. Each rate applies only to the amount inside its tier. London thresholds are £1,500,000 and £6,000,000. Taxes and any applicable resale royalty are extra. On a $10,000 hammer price, the buyer pays $12,800 (our arithmetic on published rates, before tax).

From a dealer, ask for the invoice total in writing, and ask what is outside it: tax, shipping, insurance, and whether the lab report is included or billed separately. A negotiated price that excludes three of those is not the cheaper quote.

On an online marketplace, add the platform fee, the payment fee, currency conversion, and import duty or VAT at the border. Return shipping, if the stone goes back, is usually the buyer's.

Then the costs that rarely appear in any comparison: the lab report per stone, an independent appraisal, shipping and insurance in both directions, and storage. None of them has a published standard price.

The question that separates a purchase from a position: if this stone came back to you in twelve months, at what level would you buy it?

Ask it before paying, and ask for the answer in writing. A refusal is information. A number is more information, because it states the wholesale side of the market the seller is quoting from. Most buying decisions go wrong here, not on the price of entry.

The exit has published costs on the auction route. The seller's commission at Sotheby's is a standard 10% of hammer price, with a $500 minimum on lots whose low estimate is $5,000 or less, and a 2% success fee where the hammer price exceeds the high estimate. Christie's agrees commission per consignment and may add a 2% performance commission above the high estimate. The buyer's premium paid on the way in is not recovered on the way out.

Two things that look like buy-back commitments are not. A memo, or approval, is a consignment arrangement: stones travel to a dealer to be shown, payment falls due only for stones kept or sold, and unsold stones return by the agreed deadline, under a written agreement that names each stone, its price, the deadline and the insurance. And an insurance appraisal states replacement cost at retail, meaning what it would take to buy a comparable piece. It is not an offer.

There is no reliable public figure for an average resale discount on colored stones, and none for a typical dealer margin. Treat any percentage quoted without a named source as marketing.

Where to buy gemstones is a vetting question before it is a price question. Three channels, three different documents.

Auction. Christie's states that lots are sold "as is", in the condition they are in at the time of sale. A condition report describes that state, but it is the buyer's job to request it and to read it before bidding. If a stone is offered without a report from a named lab, ask the house well before the sale whether one exists or can be obtained. There are no returns after the hammer.

Dealer or trade. This is the channel where the most information changes hands and the least of it is written down unless it is asked for. Require the treatment line on the invoice or bill of sale, not only in conversation. Trade-association membership is a weak but real signal: the American Gem Trade Association's code of ethics, for example, makes written treatment disclosure mandatory on invoices and bills of sale, and requires members to request written treatment disclosures from their own suppliers. Beyond that, ask for the return period, ask how long the business has traded, pay by a traceable method, and put any approval arrangement in writing.

Online marketplace. The listing is the weakest evidence in this guide, because photographs, video and chat are all supplied by the party selling. Two protections are structural rather than personal. The first is who holds the money, and for how long. The second is a statutory return right: in the EU, a consumer buying at a distance from a trader has 14 days to withdraw without giving a reason, with the clock starting on delivery of the goods. The public-auction exception requires that consumers can attend in person, so a purely online auction run by a trader generally keeps the right; purchases from private individuals generally fall outside it. Outside the EU, rely on the platform's written policy rather than on what a seller says in a message. Either way, a return window is the time to send the stone to a lab.

No seller is named here, because a seller's own description is not a control.

Each of these costs more the later it is found.

The lab and the report number, then the exact treatment line. A ruby described as heated with significant residues and one described as having no indications of heating are different products at different prices, and only a named lab's wording settles which one is on the table. Then compare the ask with named sales in the same size and treatment band on our ruby price per carat page.

The clarity-modification line. SSEF grades the amount of filler as minor, moderate or significant, often shortened to F1, F2 or F3, and names the filler type, while "no indications of clarity modification" means no filler was indicated in the fissures at testing. That grade records the stone's condition on the day it was tested, so a report is not a permanent description of an emerald.

The same checks apply, with two additions. Read the condition report and ask for a gemological report before the sale, and convert the estimate into an all-in figure using the premium tiers, because the hammer price is not what leaves the account. From a dealer, the equivalent step is the written invoice and the return period.

With no mount in the number, a loose stone can be compared directly against a per-carat band for its size, treatment and origin, from named sales. The most common error is comparing a fine stone's ask with a commercial stone's record, or the reverse: the two are different products, not two prices for the same thing.

This page is one: six checks, no seller names, no stone recommendations. The tests are documentary: a lab report that resolves on the lab's site, a treatment line read word by word, an origin line from a lab, a benchmark from a named sale, an all-in total, and a stated buy-back level.

Treat it as unverified and price it that way, or make the purchase conditional on a report from a named lab issued during a return window. An independent report costs money and time; so does discovering, years later, that a stone is a composite or that its origin was a story.

Every figure above comes from one of these pages, or from the Carat Brief guides linked in the text. All pages read on 11 November 2026.

Carat Brief is editorial and is not investment advice.