The Method You Cannot Choose
In most of the world the inventory costing method is an accounting preference, and any defensible one will do. In the United States it is a choice with consequences that reach past the ledger — which makes it the one market where "our system supports one method" is a fact a buyer has to hear before signing, not after. We support one method.
Every business holding stock has to answer a question that sounds academic until the first time it costs money: when you sell one unit out of a bin holding units you bought at four different prices, which price just left the building? The units are identical. The money is not.
There are three usual answers. Take the oldest cost first. Take the newest cost first. Or stop tracking individual purchases and carry one blended cost for everything in the bin, moving it each time you buy. The third is what we do, and it is the only thing we do — the merits of the first two against it are argued in oldest-first versus weighted average, which is the companion to this post and reaches a different conclusion because it is answering a different question.
For most of the world that sentence is unremarkable, and this post would not exist. A blended average is an accepted basis under international standards, it is what a great many mid-sized businesses use, and arguing about the alternatives is a pastime rather than a decision. North America is where that stops being true.
What this post will not tell you
Which method you should be on, or what your filings require. We do not publish tax positions we cannot date and source, and this one turns on your entity, your books and your adviser. What this post does tell you is a property of our software that is true regardless of your answer, and that you can confirm in a five-minute demo: there is one basis, it is not a setting, and it cannot be switched.
Why the choice has teeth in one market and not the others
The international standards used across most of the markets we serve permit the oldest-cost-first method and the blended-average method, and do not permit newest-cost-first. So in Kenya, in the Gulf, across the European Union and almost everywhere else, the newest-cost-first method is simply off the table and the remaining choice is between two options that both produce defensible numbers.
The United States is the significant exception. Newest-cost-first is permitted there and is used, deliberately, by businesses holding stock whose replacement price keeps rising — because costing sales at the newest and highest price reports a lower profit than costing them at the oldest and cheapest.
Elsewhere the costing method is an accounting preference. In one market it is an election, and an election is something you make once and then have to live inside.
That is the part that makes this a purchasing question rather than a bookkeeping one. If your accountant has told you that the method in your filings and the method in your financial statements have to agree, then your financial statements are not free to be produced on whatever basis your software happens to offer. The software has to be able to produce the method you are committed to. Ours produces one method, and it is not that one.
What we actually do, stated plainly
Every item carries one cost figure: a weighted average, recalculated from the stock on hand and what each consignment of it cost to land — freight and duty included, rather than the purchase price alone. A receipt moves that figure. A sale is costed at whatever it stands at when the sale happens. Cost of sales, margin reporting, stock valuation and the accounting integrity check all read that same figure, which is worth saying because a system using different bases in different places is a genuinely common defect and produces variances that are nothing but a difference of definition.
What does not exist is any notion of a cost layer. We do not keep a queue of the lots you bought at different prices and consume it from one end. Nothing anywhere in the product could produce an oldest-first or newest-first number even internally, because the data structure that would hold the layers was never built. This is not a setting turned off. There is no setting.
| Question a buyer should ask | Our answer | What it means for you |
|---|---|---|
| Which costing methods do you support? | One. A weighted average, maintained per item. | If your books must be on another basis, we are not your system of record for stock. |
| Where is the setting to change it? | There is not one, and there is no hidden default either. | Nothing to discover later. The answer is the same on day one and day nine hundred. |
| Can it be added for us? | It is a large build touching valuation, cost of sales and every unit-cost report — not a configuration change. | Treat it as a roadmap conversation with a real timeline, not a switch. |
| Does the same basis feed margin, valuation and the ledger? | Yes — one figure, read by all of them. | This one is a strength, and worth asking every vendor. Different bases in different screens is a common and confusing defect. |
| Can you restate the cost of stock already sold? | No. A correction is a new transaction; the average moves forward from here. | Retrospective correction of a costing error means a manual adjustment with a paper trail. |
The case for one method, made honestly
We are not apologising for this, and it would be dishonest to present it as purely a gap. A single blended basis has real properties that the alternatives do not. It cannot be gamed by choosing which lot to ship. It produces a cost figure that is stable rather than lumpy, which makes it a better input to a pricing decision than a number that jumps depending on which consignment a picker happened to reach for. And it is dramatically simpler to explain to the people who actually have to trust it.
The honest position is narrower than "one method is enough". It is this: if you are free to choose, ours is a defensible choice and a great many businesses make it deliberately. If you are not free to choose — because a commitment already exists and something else has to match it — then no argument about the merits of averaging is relevant to you, and you need a system that produces the method you are bound to.
One cost basis, maintained per item
A weighted average recalculated as stock arrives, including the freight and duty attached to each consignment.
The same basis everywhere it is read
Cost of sales posting, till margin analytics, stock valuation and the accounting integrity check all resolve to one figure, so a variance between two screens is a real variance rather than a difference of definition.
Landed cost carried into that figure
Freight, duty, insurance and handling attach to the consignment at receipt and are inside the average, so it is an arrival cost rather than a purchase price.
A choice of costing method
Not built, and not configurable. There is no setting, no default, and no internal layer structure that a different method could be computed from.
Oldest-cost-first or newest-cost-first consumption
Neither is available. This is the same absence stated the other way round.
Restating the cost of goods already sold
Not built. Recalculation moves the average forward and does not reach back into transactions already posted.
Four questions for any stock system you are evaluating
How many costing methods does the product support, and where is the setting?
What a straight answer sounds like
A number and a screen, or an admission. Ours is one, and there is no screen.
Why it matters
A vendor who says "it is configurable" and cannot show you the configuration is describing a roadmap.
Does cost of sales use the same basis as the valuation report?
What a straight answer sounds like
Yes, demonstrated on one item, in both screens.
Why it matters
Two bases produce a permanent unexplained variance between your stock value and your ledger, and it is nobody's job to find it.
Is landed cost inside the unit cost or beside it?
What a straight answer sounds like
Shown on a receipt with freight attached.
Why it matters
Freight expensed separately means every margin you read is overstated by the cost of getting the goods to you.
If I need to change method in three years, what happens?
What a straight answer sounds like
A serious answer about a migration, or a straight no.
Why it matters
Changing basis is a restatement of everything you hold. Find out before you have nine hundred days of history in it.
One thing, and we would want a reason before starting it
This is deliberately a short list, because the honest answer to "will you add costing methods" is that it is a large build and we would rather tell you that than take the order.
Cost layers, and a method that consumes them
A queue of purchase lots per item, consumed from one end, with the method selectable. It touches stock valuation, cost of sales posting, every unit-cost report and the till's margin analytics — which is why it is a project rather than a feature, and why we would want to understand what is driving the requirement before quoting it.
If you are in this position, the useful conversation is not about our roadmap. It is about whether your requirement is a genuine commitment you are bound to, or a preference inherited from whatever system you are leaving. Those need different answers and only one of them needs a build.
Tell us what your method has to beWhat AWRA OpsHub does today
- A weighted average per item, recalculated from stock on hand and what each consignment cost to land rather than from the purchase price alone.
- One basis read by every consumer of it — cost of sales, margins, valuation and the integrity check. Worth asking any vendor, because the alternative produces variances that mean nothing.
- Landed cost inside the figure, so freight and duty are part of the unit cost rather than a separate expense line.
- A stable number to price from, which is the practical argument for averaging and is not a small one.
What it does not do
- No choice of costing method. Not a default, not a setting, not a hidden preference — the concept is absent from the product.
- No cost layers of any kind, so neither oldest-first nor newest-first could be produced even internally.
- No retrospective re-costing. Correcting a cost is a new transaction, not a restatement.
- No per-location or per-category basis. One method, one organization, whatever you hold and wherever you hold it.
Not ours, by choice
- We will not tell you which method your filings require or what your statements must agree with. That is your adviser's question, it has consequences, and a software vendor answering it is doing something they are not qualified to do.
- We do not claim that a weighted average satisfies any particular jurisdiction's requirements for your business. We claim only that it is the one basis we produce.
The verdict
Almost everywhere we operate, the costing method is a question with two acceptable answers and a blended average is one of them. North America is the market where it can instead be a commitment, and a commitment is not the kind of thing you want to discover your software cannot honour after the data is in it. So we say it early and plainly: one basis, no setting, no layers underneath it. If that is fine — and for a great many businesses it genuinely is — everything else about how we cost stock is unusually consistent, and the consistency is worth more day to day than the choice would be. If it is not fine, you have learned it from a blog post rather than from your auditor.
Bring us the constraint, not the feature request
If something already commits you to a costing method, tell us what and why. We will tell you plainly whether we can be your system of record for stock, and if we cannot, we will say so before you spend anything.
Talk to us about stock costing