The four landed-cost numbers every buyer sourcing from Thailand should track.

A jewelry sourcing program lives and dies on its costing model. The buyers who survive the next five years will be the ones who replaced "unit price" — a single number from a quote sheet — with four numbers that together describe what each piece actually costs them to put on a shelf.

I have spent thirty years on both sides of the table. I have written the quote that goes to the buyer, and I have read the quote as a buyer myself. The patterns are consistent enough that I will say something blunt: the gap between the unit price on the quote and the true landed cost on the balance sheet is the single most under-managed line item in jewelry buying. It is often eight to eighteen percent — and it is invariably larger than the buyer's annual price-negotiation effort.

This essay lays out the four numbers I track for every sourcing engagement, what each one means, what most teams miss, and how to install the discipline without rebuilding the whole costing stack at once.

Why "unit price" is the wrong number

Unit price is the easiest number to negotiate, the easiest to compare across suppliers, and the easiest to put on a one-page summary for the buying meeting. It is also the easiest one to be wrong about — because it does not include the half-dozen real costs that determine whether the SKU is actually profitable.

The standard answer is "we add a landed-cost adjustment." Most operations I review do this once a year, with assumed percentages, applied uniformly across SKUs. The numbers are usually wrong in both directions: some SKUs are subsidized by the adjustment; others are penalized by it. Both kinds of error compound. By the time the next annual review happens, the wrong SKUs have been over-ordered and the right ones have been under-stocked.

The gap between unit price and true landed cost is the single most under-managed line item in jewelry buying — and it is invariably larger than the buyer's annual price-negotiation effort. — From the field

The fix is not heroic. It is to commit to tracking, by SKU, the four numbers below — and to update them at a meaningful cadence, with named ownership.

The four numbers

Number 01 — Quoted unit cost

Easy to get; usually the only one being tracked. The problem is not that the number is wrong on the day it is quoted; it is that it stops being right the moment the metal market moves, the volume tier changes, or the factory reorganizes a sub-supplier. A quoted unit cost without a date, a metal assumption, and a volume tier is a number that will mislead your costing model within six weeks.

The discipline is small. Store the quote with three modifiers: date, metal benchmark (LBMA AM/PM, or your factory's reference), and the volume tier the quote was bound to. Without these, every later comparison is fiction.

Number 02 — Effective unit cost

This is where most operations leak. The quoted unit cost assumes a yield — explicitly or implicitly. The factory has its own assumed yield, which it folds into the quote. You may not be told which yield was assumed. If you accept the quote without adjusting for your own measured yield on the actual production run, you are paying for a number that does not match what shipped.

The three yields that move materially in jewelry:

• Casting yield. How much of the metal you bought actually arrives as finished casting — typically 88–96%, depending on alloy, casting technique, and operator skill.
• Finishing yield. How much weight is lost in polishing, plating prep, and surface finishing — typically 1–3% by weight, but more for heavily worked surfaces.
• Stone-set yield. Breakage, mis-set, and reject rates on set pieces. Often the silent cost — a 2% stone breakage rate on a high-stone SKU can move effective cost by more than the labor line.

The effective unit cost re-prices the quote against your measured yields. The discipline is to measure your actual yields by production run, by SKU class — not to use the factory's assumed yields as the cost basis.

What this looks like in practice

An engagement I ran for a US wholesale brand: quoted unit cost on a popular ring SKU was $46.20. After applying their measured casting yield (92%, vs. the factory's assumed 95%), finishing yield (98.5%), and stone-set yield (97.2%), the effective unit cost was $50.84. Across a six-month run of 1,800 units, that gap was $8,352 of unbooked cost — invisible on the standard report.

Number 03 — Landed cost (true)

Effective unit cost plus freight, duty, import handling, and the documented cost of QC and rework. Most operations get freight and duty right (the numbers are external and unavoidable); most underweight the QC and rework cost.

QC and rework cost is real. Every reject inspected on receipt is an inspector hour. Every rework returned to the factory is a freight cycle, a re-inspection cost, and a delay-induced inventory cost. Track these by SKU and supplier; the variance reveals which suppliers are actually expensive when fully loaded.

Quoting is the negotiation; yield is the truth. — Practice maxim

Number 04 — Carry-adjusted cost

The number most buyers omit and most CFOs quietly worry about. Landed cost plus the cost of capital tied up — financing, insurance, security, depreciation — pro-rated by your actual cover. If a unit costs you $52 landed and sits on your shelf for nine months at a 10% blended cost of capital, you are carrying it at $55.90 by the time it sells.

You will not negotiate this number with the factory. But you will use it to decide which SKUs to buy more of and which to let go — and that decision is the one that compounds into margin over a twelve-month horizon.

Installing the discipline without rebuilding the stack

You do not need a new ERP, a new costing system, or a consultant who proposes one. You need a working spreadsheet (or a clean module in your existing system), a named owner, and a monthly cadence. The minimum viable version of this is one A4 page per SKU class — quoted, effective, landed, carry-adjusted — updated against actual production data and actual cover.

Three things make this stick:

1. One person owns the numbers. Not the buyer, not the merchandiser, not finance — one named owner who updates the model and presents it in the monthly buying review.
2. The factory sees the model. Within reason. When the factory understands that you measure effective cost, the quoting conversation changes — and your yields tend to improve too.
3. Decisions reference the model. Every reorder, every new-SKU launch, every supplier-mix decision should reference one of the four numbers explicitly. The discipline lives in the meeting agenda, not in the spreadsheet.

What to expect

Most operations that move from "quoted unit cost" tracking to the four-number framework see a six- to fourteen-percent reduction in true landed cost across the first twelve months of the program — without any change in supplier mix. The improvement comes from re-negotiating with the right number, ordering the right SKUs, and not over-ordering the ones that look cheap on a quote sheet but cost real money on a balance sheet.

The bigger gain, though, is structural. An operation that knows its true landed cost stops being surprised. The decisions get cleaner. The inventory plan reads honestly against the P&L. And the next quarter, the model is right enough that you can start asking sharper questions about which suppliers are pulling their weight.

That is the work. Less heroic than it sounds; more compounding than it looks.