What is actually inside the 25%?
Inventory carrying cost is usually quoted as a single percentage — 20%, 25%, sometimes 30% of inventory value per year — and then applied to every SKU in the building. Built from its parts for one warehouse below, it does come out at exactly 25.0%. Applied to two individual SKUs from that same warehouse, it is 51% too high on one and understates the other by more than three times. The numbers are an invented worked example, built so the arithmetic can be checked end to end.
- Carrying cost is four different costs added together: capital, storage, service and risk. They are quoted as one percentage because they are paid out of one budget, not because they behave alike.
- They do not share a driver. Capital scales with the value you hold, and service is often allocated by it. Storage exposure follows the space you occupy, but the cash cost may be fixed, stepped or usage-based, and handling labour follows movements more than space. Risk follows neither — it is a property of the SKU, and it is the component that moves most.
- Built bottom-up here: $36,000 capital + $34,000 storage + $6,000 service + $24,000 risk on $400,000 of average inventory = 25.0%.
- Split by driver, that same rate is 16.5% of value plus $1.70 per cubic foot per year — and those two pieces land on different SKUs in wildly different proportions.
- A small expensive SKU costs $6,617 a year to hold; a bulky cheap one costs $6,420. Nearly identical. The blended 25% says $10,000 and $2,000 — that the first costs five times the second.
- ⚠ And the two-driver split is itself too tidy. Change one SKU’s obsolescence risk from the warehouse average to one set by its own markdown and write-down history, and its rate goes from 16.5% to 40.5% without a single other number moving.
Ask what it costs to hold a unit for a year and you will get a percentage. Ask what is in the percentage and the conversation usually stops. That is a problem, because almost every ordering decision that claims to be economic — how much to buy, whether to take the price break, whether to hold safety stock, when to mark down — runs on this number.
It is worth building once, from the bottom, for your own building.
The four components, and what actually drives each one
The standard breakdown is not controversial. What gets skipped is the right-hand column.
| Component | What it is | What it scales with |
|---|---|---|
| Capital | The cost of the money tied up — your borrowing rate, or the return that money would earn elsewhere | Value held × time |
| Storage | Rent, utilities, racking, the handling labour you pay because stock is there to be moved | Space and rent: footprint × time. Handling: movements and touches, where measurable |
| Service | Insurance and any inventory or property taxes on held stock | Often allocated by value held × time |
| Risk | Obsolescence, markdowns taken to clear, shrinkage, damage, expiry | ⚠ The SKU itself — category, shelf life, markdown and write-down history |
Some of those components can be measured directly, others have to be allocated, and risk usually has to be estimated. We come back to risk at the end because it is the one that breaks the tidy version of this article.
Step 1 — Build the rate for the building
One warehouse, one year. Average inventory value at cost across the year is $400,000.
| Component | Working | Annual cost | As % of $400,000 |
|---|---|---|---|
| Capital | $400,000 × 9.0% blended rate | $36,000 | 9.0% |
| Storage | Rent, utilities and holding labour attributable to stock | $34,000 | 8.5% |
| Service | Insurance premium + inventory taxes | $6,000 | 1.5% |
| Risk | Write-downs, clearance markdowns, shrinkage, damage | $24,000 | 6.0% |
| Total | — | $100,000 | 25.0% |
So the rule of thumb was right, for this building, in aggregate. If the article stopped here you would have a defensible number and a completely misleading tool.
Step 2 — Split it by driver, because two of those rows are not about money
Capital scales with the value you are holding, and service and risk are often allocated by it. Storage does not — a pallet takes the same space whether it holds $200 of foam or $200,000 of components. Storage exposure follows space, but the cash cost may be fixed, stepped or usage-based depending on how the warehouse is paid for.
| Driver | Components | Annual cost | Rate |
|---|---|---|---|
| Value-driven | Capital + service + risk | $66,000 | 16.5% of inventory value |
| Volume-driven | Storage | $34,000 | $1.70 per cubic foot per year |
The $1.70 is just $34,000 spread across the 20,000 cubic feet of racking the stock occupies on average. Two numbers now instead of one, and the second is not a percentage of anything.
⚠ Treat that $1.70 as an allocation rate, not a marginal savings rate. If a 3PL bills you by pallet, bin or cubic foot, the two are close. If you pay a lease on the building, rent is largely fixed until you free enough space to shrink the footprint: drop a SKU tomorrow and you have freed capacity, but you have not necessarily saved its allocated storage in cash. For ordering and capacity decisions, consumed space still has economic value — but “this SKU costs us this much in storage cash” and “this SKU consumes this much allocated capacity” are not the same claim. Know which one your number is.
Handling labour leaks out of the space bucket too. It is driven far more by receipts, picks, relocations and touches than by cubic feet: a tiny SKU picked thousands of times can consume more handling than a bulky pallet that sits still. The example below leaves handling inside the $1.70 for simplicity; where handling is material and you can count the touches, allocate it by movement instead.
Step 3 — Put two real SKUs through it
Both from the same warehouse, both covered by that same 25%.
| SKU A — small and dear | SKU B — bulky and cheap | |
|---|---|---|
| Unit cost | $80.00 | $4.00 |
| Unit volume | 0.02 cu ft | 1.5 cu ft |
| Average units held | 500 | 2,000 |
| Average value held | $40,000 | $8,000 |
| Average space occupied | 10 cu ft | 3,000 cu ft |
Now carry them for a year:
| Cost | SKU A | SKU B |
|---|---|---|
| Value-driven @ 16.5% | $40,000 × 0.165 = $6,600 | $8,000 × 0.165 = $1,320 |
| Volume-driven @ $1.70/cu ft | 10 × $1.70 = $17 | 3,000 × $1.70 = $5,100 |
| True annual carrying cost | $6,617 | $6,420 |
| As a % of that SKU’s value | 16.5% | 80.3% |
| ⛔ What the blended 25% says | $10,000 | $2,000 |
| ⛔ Error | 51% too high | True cost is 3.2× the estimate |
⭐ Read the two totals next to each other, because that is the whole article: these two SKUs cost almost exactly the same to hold for a year — $6,617 and $6,420. The blended percentage reports that the first costs five times as much as the second.
And SKU B’s real rate is 80% of its own value per year. Hold $8,000 of it for twelve months and, on this allocation, you spend $6,420 doing so. Any price break on a bulky, cheap item has to be weighed against that number, and the 25% rule will never show it to you.
⛔ Step 4 — Where the tidy version breaks
Two drivers instead of one is a real improvement and it is still too clean, because the risk row does not belong in the value bucket. It was put there for convenience: obsolescence was 6.0% of inventory value across the warehouse, so it got folded in with capital and service and treated as though it scaled the same way.
It does not. Obsolescence is a property of the item. A stainless fastener has approximately none. A SKU with documented markdown/write-down history can carry far more risk than a stable staple. That risk does not care what your warehouse average was.
Take SKU A and suppose its history puts its obsolescence-and-markdown risk at 30% — meaning an expected annual write-down and markdown loss equal to 30% of its average inventory value — rather than the warehouse’s blended 6%. Nothing else about it changes:
| Component | Warehouse blend | SKU A, high-risk history |
|---|---|---|
| Capital | 9.0% | 9.0% |
| Service | 1.5% | 1.5% |
| Risk | 6.0% | 30.0% |
| Value-driven rate | 16.5% | 40.5% |
| Annual carrying cost | $6,617 | $16,217 |
Same unit, same shelf, same year, same rent. $6,617 or $16,217, and the difference is one component that no spreadsheet can derive for you — it comes from knowing the SKU and its history.
⭐ So the honest shape of this is not two buckets. It is that some components can be measured directly, others have to be allocated, and risk usually has to be estimated, and the one you have to estimate is frequently the largest. Anyone offering you a clean formula for carrying cost has quietly assigned your risk row an average and not mentioned it.
What the number actually informs
- Order quantity. The standard EOQ calculation has carrying cost in the denominator. Use a blended rate on a bulky cheap SKU and it will tell you to order far more than you should.
- Whether to take the price break. A higher SKU-specific carrying rate can erase a volume discount that looked attractive under the warehouse average; run the actual quantity and holding period.
- When to mark down. At a simple annualized 40% rate, six months of unchanged average exposure represents roughly 20% of value. Set that against the markdown you were avoiding.
- How much safety stock is defensible. Safety stock raises average inventory, so the carrying rate is part of deciding how much buffer is economical.
- What it should not decide: anything, at the blended rate, for a SKU whose value density or risk profile is nothing like the warehouse average. That is most of what the blended rate is used for.
What to do this week
- Add up the four components once, for your building. Your actual capital rate, your actual rent and holding labour, your actual insurance, your actual write-downs. It will be better than the rule of thumb.
- Split it into a value rate and a space rate. Divide storage costs by the cubic feet your stock occupies, and note whether that storage cash is fixed, stepped or usage-based. Where handling labour is material, allocate it by touches rather than space. This is the step that changes decisions.
- Run your bulkiest cheap SKU through both. That is where the blended rate is most wrong, and it is usually a SKU nobody has questioned.
- Set risk by category, not by average. Start with a simple risk classification your team can actually maintain—low/normal/high if that is enough for your catalogue—and refine where the dollars justify it.
- Write down what went into the number and date it. Rates move; a carrying cost percentage with no derivation behind it becomes folklore within a year.
Questions people ask
What is inventory carrying cost?
The total annual cost of holding stock, usually expressed as a percentage of average inventory value. It has four components: the cost of the capital tied up, storage, service costs such as insurance and inventory taxes, and risk costs such as obsolescence, markdowns, shrinkage and damage. They are quoted as one percentage because they come out of one budget, not because they behave alike.
Is 25 percent a good estimate for carrying cost?
It is a reasonable starting point for a whole building and a poor one for any individual SKU. In the worked example above a warehouse built from its four components does come to exactly 25.0 percent, yet applying that figure to two SKUs inside it overstates one by 51 percent and understates the other by more than three times. The aggregate is not wrong; applying it uniformly is.
How do I calculate carrying cost for a single SKU?
Split the rate by driver first. Capital scales with the value held, and service and risk are often allocated by it, giving a percentage: 16.5 percent in the example. Storage is allocated by the space occupied, giving a rate per cubic foot per year: $1.70 in the example. That is an allocation rate, not necessarily cash you would save by dropping the SKU, and handling labour may be better allocated by movements than by space. Apply both to the SKU. A small expensive item is almost entirely value-driven; a bulky cheap item is mostly volume-driven and can exceed 80 percent of its own value per year.
What is the biggest component of carrying cost?
It depends on the item, which is the point. Across the whole warehouse in the example, capital is the largest at 9 percent of value. For a bulky low-value SKU, storage dominates everything else. For a SKU with a documented markdown or write-down history, obsolescence risk can be larger than the other three combined: moving one SKU’s risk from a 6 percent warehouse average to an expected annual write-down and markdown loss of 30 percent of its value takes its rate from 16.5 percent to 40.5 percent with no other number changing.
Should carrying cost include the cost of capital?
Yes. Money in stock is money not available for anything else, so the relevant rate is either what you pay to borrow or what the capital would earn in its next best use, whichever genuinely applies. Leaving it out is common and understates the rate substantially: it is 9 of the 25 percentage points in the example.
Why does carrying cost matter for order quantity?
Because it sits in the denominator of the standard EOQ calculation, so a rate that is too low recommends larger orders than the business can justify. It also informs whether a supplier’s volume discount is worth taking: a higher SKU-specific carrying rate can erase a volume discount that looked attractive under the warehouse average; run the actual quantity and holding period.