Auction pricing,explained properly.
There is no product to demonstrate yet, so here is the next most useful thing: how the numbers around an auction actually behave, and what we intend to do about them. If you price jewellery for a living, none of this will be new. If you sign off on someone else’s figure, some of it might be.
Every panel below illustrates a mechanism. None contains a value, and none is drawn from a real sale.
Last updated .
What’s the difference between a hammer price and an asking price?
An asking price is an aspiration. A pre-sale estimate is a forecast, made by the house, before anyone has bid. A hammer price is what somebody actually paid on the day. The three are routinely treated as interchangeable, and the confusion is expensive.
The estimate deserves particular care. It is produced for a purpose: to attract bidding, and to set expectations with the consignor. That purpose is not the same as describing worth. It is a reasonable professional guess, made in advance, by someone with an interest in the outcome. It is not a record of anything.
Only the hammer has a buyer attached. It is the one figure of the three that can be treated as evidence, and it is where everything else here starts.
Whether you can see the hammer at all depends on the house, and sometimes on the format of the sale rather than its age: does Bukowskis publish its hammer prices? walks through one house where the folklore had the mechanism wrong.
What is a buyer’s premium, and what does it add to the hammer price?
On top of the hammer sits a buyer’s premium: a share of the winning bid, taken by the house, usually tiered so that the share changes as the hammer rises. On top of that come fixed fees, tax where it applies, and getting the piece home.
None of this is hidden. Every house publishes its terms. The difficulty is that no two ladders are alike, they change without much ceremony, and they apply differently to the same lot depending on where it sells. Comparing a hammer at one house with a hammer at another, without accounting for what sat on top of each, is comparing two different things.
The practical consequence is specific: the number a client remembers is the total that left their account, and the number in the record is the hammer. Reconciling those two afterwards is where valuations quietly go wrong.
A published example, since we quote no rates of our own: Bonhams sets out its ladder separately for the United Kingdom and the United States. One house, one kind of lot, two different arithmetics. Read the terms of the house that actually sold the piece.
One comparable sale is an anecdote.
A single result tells you what one buyer paid on one afternoon. Two people who both badly wanted the piece will produce a figure no other buyer would have matched. One tired room will produce the opposite. Neither is wrong, and neither is a valuation.
What makes a set of results meaningful is not that there are more of them. It is whether the pieces are genuinely alike in the ways that move price:
- The same kind of object in the sense a specialist would accept, not merely the same category.
- Comparable condition and completeness. A missing element is not a discount; it is a different object.
- Provenance and documentation, where those change what a buyer will pay.
- Sold in a comparable market, in a comparable period. Taste moves, and it does not move uniformly.
When those hold across enough results, the spread between them stops being noise and starts being information. That spread is what a band is made of. The outliers are not errors to delete. Sometimes they are the most informative results in the set, and deciding which is which is the work.
This is the market approach, as the profession already defines it rather than as we would like to define it: see the International Valuation Standards, where IVS 103 sets out the valuation approaches, and RICS Valuation – Global Standards. Volume on its own has been tried as a business, and recently: what happened to Barnebys covers the auction-search group whose parent went into bankruptcy this spring, and what that means for anyone who prices lots.
Why is a price range better than a single valuation figure?
A point estimate invites an argument about the last digit, and it is an argument nobody can win. A range says something more useful and more honest: here is where the evidence puts this, and here is how firmly.
Width carries the meaning. A narrow band says comparable results are plentiful and consistent. A wide one says they are sparse, scattered, or old. It tells the person relying on it how much room there is to disagree before anyone is being unreasonable. Two valuations can be equally correct and still differ completely in how much weight they will bear.
This matters most in the moment a valuation is challenged. “It is worth this” invites a counter-figure. “The evidence puts it in this range, and here is why the range is this wide” invites a conversation about the evidence, which is a conversation you can have.
Not a hedge, and not our invention. The IVSC’s 2026 perspectives paper on managing and communicating value uncertainty makes the same argument: a rigorous, standards-compliant valuation can still yield a range of credible outcomes, and disclosing that strengthens confidence in it rather than weakening it.
Per lot, or per item. Say which.
Jewellery arrives in groups: a tray of rings, a parcel of loose stones, a box of watches sold together. Any figure attached to a group has to state whether it describes the group or one thing in it, and the two are wildly different quantities.
This is not pedantry. It is the most common way a valuation of grouped pieces goes wrong, and it becomes invisible the moment the figure is copied into a schedule. Nothing about a number says which of the two it was. We got it wrong ourselves once, which is why the distinction is now written into how the method describes anything about a group, rather than left to the reader to infer.
Which company a result came from needs the same discipline as the figure itself. Houses fail and reappear under near-identical names, and a file that points at the wrong one is how a document gets taken apart: is Campen still operating? is a current example.
What we will refuse to answer.
Stating the limits is the most credible thing available to us, and the least expensive. A method that always produces an answer is not confident. It is indifferent.
Where the evidence will not carry a reading, the answer is that we cannot say.
- Where comparable results are too few, too old, or too unlike the piece in front of us.
- Where the description does not settle what the object actually is, and a guess would decide the answer.
- Where the band would be so wide that quoting it would mislead more than it informs.
- Where the real question is authenticity or condition. Those need a specialist with the piece in their hands, and no amount of price data substitutes for one.
And one thing we will not claim: the method has not yet been tested against results held back for the purpose. Until that has been done and the outcome is known, nothing here says or implies that it is accurate, proven, or better than the estimate an auction house would produce. When there is a result worth reporting, it will be reported plainly, including if it disappoints.
If you would rather read about how we work than how the market works, that is on the about page.
There is nothing to try yet. You can be first when there is.
Leave an address and we will write once, when there is something to see. If you value jewellery professionally and have a live case, write to us instead. Those conversations shape what gets built.
Or write to marcusclay@lotwise.eu.