Stockout vs overstock: which cost should set your buffer
A stockout costs repeat revenue you never see again; overstock ties up cash for months. Here is how to size a reorder buffer per SKU.

You are staring at a reorder form at 11pm, wondering whether to add a few more units just in case. That decision, repeated across every SKU you carry, quietly sets how much cash sits in stock and how often you disappoint customers. Both sides of that decision cost money. The stockout costs contribution you never book, plus the scramble to reorder. The overstock costs cash you cannot spend and markdown risk you carry for months.
Here is the part most safety-stock advice skips: the two costs are not symmetric. The same buffer that protects a hero SKU can quietly tie up cash on a long-tail line that sells a handful of units a month. We will put a number on both sides, then show you how to size the buffer per SKU against that asymmetry instead of spreading one safety-stock percentage across the catalogue. You will see what a stockout actually costs, what overstock actually costs, the two numbers you need per SKU, and a worked example with one hero SKU and one long tail SKU. We will also cover what the buffer does to your cash plan and what to do before your next reorder.
The short answer: the bigger cost sets the buffer
A stockout and an overstock are both cash costs, but they are not symmetrical, so one safety stock percentage applied across the catalogue cannot be right. Hold more where being short costs more than being long, and hold lean where it does not.
The asymmetry in one line: a stockout on a hero SKU takes repeat revenue you never see again, while an overstock on a long tail SKU takes cash you cannot spend for months.
IHL Group's Inventory Distortion Study, reported by Chain Store Age at $1.77 trillion in global retail losses, is context only. It says the problem is large, not what your buffer should be.
What this post hands you: a cost per unit for being short, a cost per unit for being long, and a ratio that turns both into a buffer size you can defend. No forecasting software or statistics background needed: two numbers per SKU, a spreadsheet, and a reorder point calculator if you would rather the arithmetic was done for you.
What a stockout actually costs
A stockout is not only an empty shelf. It is stock that will not cover expected demand before the next delivery lands, so the risk window is your lead time, not this morning's count. Track the runway with days inventory outstanding.
Count contribution, not revenue. A missed sale costs the margin, not the price tag; price times lost units overstates the damage.
The bills come anyway: expedited freight, split shipments, a rush buy at a worse cost price, staff chasing stock. Real cash out, nothing to do with margin.
Worst is repeat revenue, the loss with no report line. Frequency beats severity: a weekly miss is a big miss by December.
What overstock actually costs
Every extra unit is cash you've already spent, money that can't go into stock that sells or marketing that brings buyers. That's the whole cost, and it's immediate.
Carrying cost is yours to fill in, per year: cost of capital, storage, insurance, shrink, obsolescence. This post has no industry average to quote, so set your own rate and write it down.
Markdown risk costs margin, not just shelf space, and long tail lines carry most of it because they're last to sell at full price. That drag shows up in your inventory turnover ratio long before the markdown does.
The tell is your bank balance. Overstock never shows as a loss in any one month. It shows as a balance that won't cover payroll while the stockroom is full.
Why the two costs are not symmetric
Overstock is measurable and feels bigger; often it is not. Hero SKUs carry repeat revenue and habit that rebuilds elsewhere; long-tail overstock is cash drag, disproportionately large versus revenue because those lines sell slowly, rarely repeat, and are marked down first. One catalogue safety stock percentage overprotects the tail and underprotects the hero. Stockout cost is partly estimated: lost sales are never recorded. Overstock cost is measurable from invoices. Don't let the paper trail make the measurable side look bigger. That gap is why a reorder point set per SKU beats a blanket percentage.
| Cost type | Hits | When | Measure |
|---|---|---|---|
| Stockout, hero | Repeat revenue, habit | Months later | Estimated lost contribution |
| Stockout, long tail | One sale | Same week | Days out x daily sales |
| Overstock, hero | Cash, some markdown | Quarter end | Invoice plus carrying rate |
| Overstock, long tail | Cash, deep markdown | Months later | Invoice plus markdown |
The two numbers you need per SKU
The fix is to put two figures on one row per SKU, both per unit: what it costs to be one unit short, and what it costs to hold one unit for a year. Per unit is what makes the comparison fair, because it strips out order quantity and season length, so a hero line and a long tail line sit on the same spreadsheet and can be read side by side.
Count only costs that move with the decision. The warehouse lease you pay either way is not part of holding one more unit; the interest on the cash inside it is.
Write every assumption next to its number. A stated assumption you can argue with beats a borrowed average nobody can trace.
Cost of being one unit short
With those two per unit figures in place, start here:
Cu = contribution per unit x share of that demand you never recover
The unrecovered share is the judgement call. A hero SKU with no substitute sits high; a long tail line where buyers switch within your catalogue or wait sits low, because the sale moves rather than disappears. Add scramble cost per unit: expedited freight on the rush order divided by units on that order. Keep it separate from margin so you can see which half hurts. A usable starting shape is (expected daily sales minus actual sales) x price x days out, but swap price for contribution before you use it. Label any output illustrative. Backorders turn the loss into delay cost and churn risk, so do not apply the same Cu to made to order lines. For how extra cover hits cash, see this 8-week cash flow forecast example.
Cost of holding one unit for a year
Now the other side: what one unit costs you while it waits.
Co = unit cost × annual carrying rate + expected markdown per unit
Set the carrying rate yourself from cost of capital, storage, insurance, shrink and obsolescence. State the year it reflects, because a rate built when borrowing was cheap is out of date now.
Expected markdown = markdown probability × retail price × discount depth
A 30% chance of clearing at 40% off a £35 item is about £4.20 of margin gone per unit.
Match the periods before comparing. If Co is annual and Cu covers one missed sale, convert one first, or the ratio lies.
Do not add the full unit cost to Co. That cash comes back when the unit sells; carrying cost is the rent you pay while it waits.
Turning the two numbers into a buffer size
With Cu and Co matched, the balancing point is the critical ratio: the share of time you want demand covered.
target in stock probability = Cu / (Cu + Co)
If being short costs far more than holding, the ratio sits near 1, so hold thick and accept the cash cost. If holding costs more, it falls below half, so hold lean and accept some demand going elsewhere.
The ratio sets direction and protection, not a forecast. Your sales history and lead time supply the units. A catalogue-wide safety stock percentage gives hero and long tail SKUs the same target, like one tax rate across every product line.
Treat this as a two number rule of thumb. Sanity check current cover, then adjust for seasonality, contracts or retail listings that change the cost of being short. Check the cash with the stock order affordability calculator.
A worked example: one hero SKU, one long tail SKU
Apply it. Figures are illustrative, not a case study; swap in your own price, cost and lead time.
Hero: 300 units/month, £40 price, £16 cost, £24 contribution; 60% of unmet demand lost. Cu = 24 × 0.60 = £14.40 per unit short. Carrying: £16 × 24% = £3.84; markdown: 0.05 × 40 × 0.40 = £0.80; Co = £4.64. Ratio = 14.40 / (14.40 + 4.64) = 0.76, so cover 76% of demand.
Long tail: 20 units/month, £35 price, £14 cost, £21 contribution; 30% lost. Cu = 21 × 0.30 = £6.30. Carrying: £14 × 24% = £3.36; markdown: 0.30 × 35 × 0.40 = £4.20; Co = £7.56. Ratio = 6.30 / (6.30 + 7.56) = 0.45, so cover 45% of demand.
Hero justifies weeks of extra cover; tail almost none, with smaller orders. Same catalogue, opposite answers, both right. Convert to units with your sales history. Highest buffer where a stockout costs a customer, not a sale.
What the buffer costs your cash plan
Once you have that ratio, translate it into units, then into cash. Price the buffer before you agree to it: units per week of cover multiplied by unit cost. At 70 units a week on the £16 hero SKU, one extra week of cover is £1,120 out of the bank before a single sale. That is the number to hold against the ratio, not the percentage.
Rank SKUs by the ratio, not by revenue, then set buffer by group: hero, seasonal, long tail. Three groups with a stated assumption each is enough for most catalogues. Revisit quarterly and after every markdown and every stockout, because both sides of the ratio move and the side that moved last is the one people argue about.
This is where Cushion fits, after the method rather than before it: one place where the cash forecast sees the stock and the stock plan sees the bank, reading bills, balances and payment terms across read-only connections to Shopify, Xero and QuickBooks. If you want the numbers done for you, start with the reorder point calculator, and for the wider method see inventory planning that aligns stock with cash. Join the waitlist.
What to do before your next reorder
Before your next order goes out, do four things, in this order.
Rank your top 20 SKUs by the ratio of short cost to holding cost, not revenue. That ranking is the only thing worth changing before your next reorder.
Write down the two assumptions that move the answer most: how much unmet demand you truly lose, and your carrying rate. The rest is arithmetic.
Set the buffer by group. Thick on hero lines, thin on the long tail, and say why in the note column.
If the stock plan and the cash forecast still live in two spreadsheets that disagree, fix that first. A finer buffer you cannot fund on the day is no improvement.
FAQs
Is a bigger buffer always safer?
No. Each extra unit is cash out and carrying cost in. Past the balancing point you fund protection the ratio says you will not use.
Is a stockout always more expensive than an overstock?
No. A stockout costs more where repeat revenue and habit sit behind the SKU, because the customer rebuilds that habit elsewhere. It costs less where buyers substitute, wait, or accept a backorder. The ratio settles it per SKU: when the cost of holding a unit exceeds the cost of being short of one, being short is the cheaper failure.
Can I keep using one safety stock percentage across the catalogue?
You can, but a single percentage prices every SKU as though being short and being long cost the same. That is exactly the assumption this method exists to test.
A sensible middle step is two or three groups, each with its own safety stock percentage. Choose those percentages from the ratio bands you have calculated, not from habit. Hero, seasonal and long tail is enough for most catalogues.
How do I measure demand I never recorded because the item was out?
To measure that unrecorded demand, take the SKU's average daily sales from the weeks either side of the gap, multiply by days out of stock, then apply the share you believe you never recover. As a starting shape, adjusted for lost rather than recorded sales:
Unrecorded demand = average daily sales x days out of stock x unrecovered share
Keep that share stated, not guessed silently. Then check repeat orders. If customers kept buying elsewhere in the catalogue but stopped on this SKU during the gap, that difference is repeat revenue the stockout took. Add it as a separate signal, not extra units on the same line.
Conclusion
The cost that sets your buffer is the bigger one, and it is rarely the same across your catalogue. Stockouts cost more where repeat revenue and habit sit behind a SKU, because customers rebuild that habit elsewhere. Overstock costs more where cash is tight and holding costs quietly accumulate. The two costs are not symmetric, so a single safety stock percentage prices every SKU as though they were. Work out the cost of being one unit short and the cost of holding one unit for a year, then let the ratio settle the buffer size per SKU or per group.
Your next step is to calculate those two numbers for your top twenty SKUs by revenue. Start with the hero items where a stockout hurts most. Cushion brings the cash forecast and the stock plan into one view, so the buffer you choose shows up in both, with read-only connections to Shopify, Xero and QuickBooks. Join the waitlist when you are ready. Price the two failures honestly, and the buffer stops being a guess.



