Quotes that are not like-for-like: six differences and the arithmetic that makes them one figure

Two quotes arrive for the same part. One is $4.20 a metre delivered; the other is €3.40 a metre, and the second looks like the better offer by a distance. It is not the better offer. The €3.40 stops in a factory yard in another country, before freight, before duty, before a clearing agent — and its thirty-day terms are the domestic supplier’s thirty days exactly, which is to say the one difference a buyer usually counts has already cancelled itself. By the time the rest is inside one column the 13 per cent advantage has become a 1.2 per cent disadvantage, and the buyer who compared the two unit prices compared two numbers that were never prices of the same thing.

This is the whole of the discipline of normalisation, and it is arithmetic rather than judgement. A quotation is a price at a point: at a place, at a time, on stated terms, for a stated quantity, in a stated currency. Two quotations are comparable only when both have been carried to the same point. Carrying them is six conversions, each of which a buyer can write down and a reader can check, and most of the failures in the work come from doing five of them and stopping.

The piece sets out the six differences, the order the conversions go in, a worked example that carries an import and a domestic quote to one basis line by line, the point at which normalisation stops being able to help, and the one decision it should change.

The six differences

Each of the six has a tell on the page and a conversion that turns it into a number. Table 1 is the working list; the sections that follow take them in the order the arithmetic requires, because three of them cannot be converted until the ones before them are settled.

Table 1. The six differences between quotations that are not like-for-like, where each shows up on the document, and the conversion that removes it. Source: the studio’s purchase manager specification, which normalises every quotation to one stated basis before ranking anything.

The first three are mechanical and the second three are judgement, and the split matters because the mechanical ones are where most of the money is. Freight and duty on an import are frequently ten to fifteen per cent of the goods value, and they are visible in the quotation if anybody reads the lines below the total. Payment terms and warranty are worth single-digit percentages and are invisible unless the buyer goes looking.

The order the conversions go in

Currency first, because every later step is in the buyer’s money. One rate, one date, one source, stated in the table. Where a supplier quotes in a currency the buyer does not hold, the exposure is real and belongs in the comparison as a line rather than in a treasury conversation afterwards.

Delivery term second, because it decides what is already inside the figure. Ex-works means the price stops at the factory gate and the buyer pays the journey; delivered duty paid means it stops at the site and the journey has been priced by someone else, at their margin. Between those two ends sit a dozen incoterms, and each one moves a cost and a risk from one party to the other. The comparison needs one end: the buyer’s site.

Tax third, and the distinction here is the one that produces the largest phantom differences. Recoverable value-added tax is not a cost to a registered buyer: it is paid, reclaimed, and irrelevant to the choice. Duty and non-recoverable tax are costs, they land in the landed figure, and they are frequently omitted because they are not on the supplier’s quote at all — they are on the customs entry, which arrives later, to somebody else, in another department.

Quantity fourth, because a price per unit is meaningless without the quantity it was quoted against. A supplier quoting 500 units at one rate and 400 at another has quoted two offers, and comparing the first against a competitor’s 400-unit price is not an arithmetic error — it is a different question. Where a quantity break is genuinely available, it goes in the table as a separate scenario with the extra units priced at their marginal rate, and the reader decides whether the extra stock will be consumed.

Payment fifth, and it is small enough that buyers skip it and large enough to matter when one supplier offers thirty days and another sixty. Thirty days of working capital on eight thousand dollars at an eight per cent cost of funds is about fifty-five dollars, or two thirds of one per cent of the goods value. In the worked example below both quotes sit on thirty days against a sixty-day standard, so the conversion moves both figures by almost the same amount and settles nothing between them — which is worth seeing, because that is the normal case, and it is why the line gets skipped even where it matters.

Warranty and lead time last, because each is a judgement with an arithmetic core. The warranty gap can be priced if the organisation has a stated price for replacement cover; if it does not, the gap is stated and not monetised, and the table says which. The lead time converts into stock: an imported item that takes 34 days to arrive forces a buffer that a 7-day item does not, and the buffer’s value carries a cost of funds for as long as it is held.

Worked example: 2,000 metres of cable

The organisation in this example is illustrative, and its policies are stated rather than assumed: 60-day payment terms, a cost of funds of 8 per cent a year, a 24-month warranty as the standard for this class of goods, three weeks of usage held as buffer on imported items, and 8,000 metres of annual usage. The quotes are constructed for this piece. Table 2 puts the two quotations side by side as they arrived; Table 3 carries both to the one basis, with the arithmetic in the last column.

Table 2. Two quotations for 2,000 metres of cable, as they arrived. Constructed for this piece; not a real supplier file.

Table 3. The two quotes carried to one basis: delivered to site, duty and clearing inside the figure, recoverable VAT outside it, 60-day payment, 24-month warranty, and the stock the lead time forces.

Five of the six conversions moved a number, and the fourth did not: both quotes are for 2,000 metres, so there is nothing to convert. Every line above the total can be checked from the two documents, which is the test the table has to pass.

The ranking inverts on the duty line. On the quoted unit price Kowalski is 13.1 per cent cheaper — $3.6516 a metre against $4.20. On landed cost Kowalski is dearer by $74.35, which is just under one per cent of the domestic price, because $601.44 of freight, $474.71 of duty and $95.00 of clearing have consumed the entire advantage and $74.35 besides. On the full basis Kowalski is dearer by $105.29, or 1.2 per cent. Nothing about the second quote changed between those three readings. What changed is where the price stopped.

Two further points come out of the arithmetic and neither is in the total.

The first is exchange risk, and it is larger than the gap. Kowalski’s price is in euros, and a two per cent move against the dollar on the €6,800 goods value is $146.06 — more than the $105.29 the normalised comparison is arguing about. Either the buyer fixes the rate with a forward contract, which costs something, or the comparison carries a scenario line for a two per cent adverse move and the reader sees that the decision is inside the noise. A comparison that treats a foreign-currency quote as a fixed number has taken a position on a market it has not priced.

The second is the buffer, and it is the line most often left out. A 34-day lead time against a 7-day one is 27 working days, and the organisation’s stated policy is to hold three weeks of usage against it: 8,000 metres a year is 461.54 metres in three weeks, worth $1,955.62 at the landed cost of $4.2372 a metre, and 8 per cent of that is $156.45 a year for as long as the item is imported. That line is the reason the normalised gap is wider than the landed one: the cheap quote has bought freight with distance and time with stock, and the stock is not free.

The honest reading of the finished table is that the two offers are within 1.3 per cent of one another, in the opposite order to the one the unit prices suggested. A 1.3 per cent gap is smaller than the largest unmonetised difference between them, which is a 24-month warranty against a 12-month one, and smaller than the exchange exposure the import carries. The comparison has therefore done its job precisely by failing to produce a winner: it has replaced a false 13 per cent with a true 1.3 per cent, and moved the decision from price to terms.

Where normalisation stops

Normalisation converts what a document states and cannot convert what it does not. Quality is the clearest case: if the domestic supplier’s cable is on the approved list and the import’s is not, the comparison is between a known material and an unknown one, and no line in the table holds that. The same applies to capacity — a supplier who can ship 2,000 metres this month and 20,000 next quarter is not the same supplier as one who cannot, at any price — and to service, which is what a buyer is actually buying when the item fails.

The second limit is the assumptions, and they are the buyer’s own. This table is built on an 8 per cent cost of funds, a 1.5 per cent annual price for warranty cover, and a three-week buffer policy. None of those is a market observation; each is a figure the illustrative organisation stated. Change the cost of funds to 12 per cent and the payment line grows by half; drop the buffer policy and $156.45 disappears, at which point Kowalski is cheaper again by $51.16. That sensitivity is not a flaw in the method. It is the method working: a conclusion that flips on a stated assumption is a conclusion the reader can argue with, which is a great deal better than one that cannot be examined at all.

The third limit is the one that costs money. A normalised table looks authoritative — it has decimal places, a rate, a date and a total — and an approver who signs it has not seen the two quotations. Where the extraction was wrong, the table is confidently wrong, and the error is invisible in a document that is internally consistent. The read-back against the source documents is not a courtesy; it is the only control on the conversion, and it is one call.

The fourth limit is time. A quote is normalised on the day it is read, at the rate and the terms then in force. Duty rates change, freight moves, and a quote with 21 days’ validity may not survive the approval it is meant to support. The table carries the date for that reason, and a comparison older than its shortest validity is a historical document.

The cost

Reading two quotes in two layouts is two calls; building the normalised table is one; reading it back against the source documents is one. Four calls at the measured Base rate of $0.00700 is $0.02800 for the pair. A cycle with three quotes is five calls, $0.03500; sixty such cycles is the 300 calls and $2.10000 that the purchase manager’s own page assumes.

The marginal figure is worth stating separately, because it is the one that decides whether a buyer bothers. The third quote’s extraction is one call, $0.00700 — and the third quote’s four conversions are inside the same call that builds the table. The expense of the third quote is not the model. It is the days spent waiting for it, and the days are what a quote-count policy is actually buying.

The decision

Normalise before comparing, and state the basis in the table — the rate and its date, the delivery point, the tax treatment, the payment term, the warranty standard and the quantity. A comparison whose basis is stated can be argued with, re-checked and defended a year later by someone who was not there. One that is not stated cannot be defended at all, and the person who has to defend it will be the buyer.

Then convert everything that can be converted and name what cannot. Quality, capacity, service and exchange exposure do not disappear because they have no column; they become sentences immediately under the total, where the approver reads them before signing. A table that quietly omits them has made the decision sound cheaper than it is.

And apply one rule at the end: where the normalised gap is narrower than the largest term that could not be normalised, price is not the decision. Two offers within 1.3 per cent, one carrying twice the warranty and half the lead time, are not separated by 1.3 per cent — they are separated by the warranty and the lead time, and the comparison’s job is to say so rather than to pick the smaller number.

Where this goes

• The pillar: x402.review/2026/10/31/commercial-operations/ — one record, two readers, and the approvals between them
• The role that does the arithmetic: studio.x402.press/roles/purchase_manager — the comparison on one basis, the approval record, the order and the match
• The side that drafts the quote: studio.x402.press/roles/sales — the proposal, the terms sheet and the figures it may not invent
• The service it buys: x402.press/services/deepseek/ — the canonical price, seven chains
• The review: x402.review/2026/10/31/deepseek-review/ — the measured handshake and what it does not promise

Method note

The price was read from the live 402 challenge on Base on 31 October 2026, outside DeepSeek’s weekday peak windows of 01:00–04:00 and 06:00–10:00 UTC: 7000 atomic units under exact, which is $0.00700, and 9550 under upto at the max_tokens tested. Call counts are planning assumptions, stated in the cost section; four calls for a two-quote cycle and five for three reconcile with the purchase manager page’s 300 calls for sixty cycles.

The suppliers, the rates, the freight and duty figures and the organisation’s stated policies are constructed for this piece and labelled as such. The 1.0740 rate is an illustrative rate, not a rate observed on the date of writing, and no shipment, customs entry or purchase described here took place. The cost of funds, the warranty replacement price and the buffer policy are the illustrative organisation’s own stated figures; a reader who substitutes their own will get a different arithmetic from the same two quotes, which is why each line shows its working.

x402.review is written and edited by N.P. Vincent. The publication is connected to x402.press, which sells the services reviewed here; that interest is disclosed on every piece that touches a service, and it is stated here rather than left to inference.

Filed under Jobs · Tags: procurement, quotes, normalisation, comparison