TL;DR & Quick Summary
Most stockout calculations multiply missed units by price and stop there. That overstates the loss badly, because it assumes every unfilled unit is a permanently lost sale — and understates it in the cases that matter most, because it ignores customers who leave and do not come back.
This is the method for producing a number you can actually act on.
The formula:
Unfilled demand × gross margin per unit × true loss rate + downstream customer loss
Where each input comes from — your own systems, not a benchmark table
The input everyone skips: what customers actually do when the shelf is empty
Benchmark out-of-stock rates to sanity-check your own measurement
Key Takeaway: The number is rarely as large as vendor calculators claim and rarely as small as finance assumes. It varies enormously by product, and that variation is the whole point.
Get Started: Want this run against your own stock and sales history? Schedule a Strategy Call with Cogniq AI or explore our predictive analytics services.
Why the Simple Version Is Wrong
The calculation most people run is:
missed units × selling price = lost revenue
Three things are wrong with it.
It uses price, not margin. A stockout does not cost you the sale value. It costs you the contribution you would have earned on it. Using revenue inflates the figure by whatever your cost of goods is — frequently by a factor of two or more.
It assumes every missed unit is lost. Many customers wait, backorder, or return tomorrow. Those sales are delayed, not lost. Counting them as losses is the single largest source of inflation in vendor-supplied calculators.
It ignores what happens next. Conversely, some customers do not just skip the purchase — they buy from someone else and stay there. That is worth more than one unit of margin, and the simple formula cannot see it at all.
The result is a number that is simultaneously too big on your fast-moving staples and too small on the products where a stockout genuinely does damage.
The Four Inputs
Input 1: Unfilled demand
Where to get it: your inventory system's out-of-stock records, cross-referenced with demand history for the same SKU in comparable in-stock periods.
This is harder than it looks, because you cannot observe demand you failed to serve. Nobody records the customer who looked and left. Two workable approaches:
- Comparable-period estimation. Take average daily demand for that SKU in recent in-stock periods, multiply by days out of stock. Crude, but defensible.
- Same-store, same-week comparison. For multi-location businesses, compare a location that stocked out against one that did not, over the same week. Better, because it controls for seasonality and promotions.
Adjust for trend and season. A line that stocks out during its peak week loses far more than its annual average suggests.
Input 2: Gross margin per unit
Where to get it: your accounts, not your price list. Selling price minus cost of goods. If margin varies by channel, use a weighted average across the channels the SKU actually sells through.
Input 3: True loss rate
Where to get it: this is the one you have to reason about rather than look up, and it is the input that decides whether your answer is credible.
Published behaviour research gives useful anchors. Around 70% of shoppers may buy a different brand when their usual choice is unavailable, roughly 30% may visit another store, and about 43% will go to a competitor (NetSuite, Opensend).
But those are averages across all categories, and your situation is more specific:
| Customer behaviour | What it costs you |
|---|---|
| Waits and buys later | Almost nothing — a little goodwill |
| Buys your alternative product | The margin difference between the two |
| Buys a competing brand you also stock | The margin difference, often small |
| Goes to a competitor | The full margin |
| Goes to a competitor and stays | Full margin plus future value |
Estimate the split for the product in question. A staple with no shelf alternative skews heavily toward "waits". A discretionary item beside three substitutes skews heavily toward "switches".
Use a range, not a point estimate. If you think 40–60% is genuinely lost, calculate both ends and present the range. A single confident number here is false precision.
Input 4: Downstream customer loss
Where to get it: your repeat-purchase data.
Apply only where a stockout plausibly ends a relationship — subscription products, regularly repeated purchases, B2B supply agreements. For an occasional discretionary buy, set this to zero and say so.
Where it does apply, use realised repeat value, not lifetime value projections. Lifetime value is where these calculations become fiction.
Worked Example
Illustrative arithmetic at stated assumptions, not client data.
A distributor examines one high-margin SKU over a quarter:
| Input | Value | Source |
|---|---|---|
| Days out of stock | 14 | Inventory system |
| Average daily demand when in stock | 12 units | Sales history |
| Unfilled demand | 168 units | 14 × 12 |
| Gross margin per unit | £34 | Accounts |
| True loss rate | 45–65% | Estimated: 3 shelf substitutes |
| Downstream loss | £0 | Occasional purchase, no subscription |
168 × £34 × 0.45 = £2,570 168 × £34 × 0.65 = £3,713
Quarterly cost: £2,570 – £3,713 on this SKU alone.
Compare against the naïve calculation: 168 × £89 selling price = £14,952. The simple method overstates by roughly four times.
Now repeat across the SKUs that actually stock out. The ranking that comes out of this exercise is more useful than the total, because it tells you where safety stock earns its keep.
The Stockouts Your System Cannot See
Everything above assumes your inventory records are correct. For most businesses they are not, and the gap creates a category of stockout that never appears in any report.
Phantom stock is inventory the system believes you hold and you do not. The record says twelve; the shelf has none. The causes are mundane and universal: mis-scanned receipts, unrecorded shrinkage, damage written off late, returns processed to the wrong location, units sitting in the wrong bin.
This is the most damaging failure mode in inventory management, because it is silent by construction. A visible stockout at least triggers a reorder. Phantom stock suppresses the reorder — the system sees available inventory, so no replenishment fires, and the shelf stays empty until someone physically notices. That extends the outage from days to weeks, and it means your measured out-of-stock rate understates reality.
Three ways to find it without a full stock count:
Watch for zero-sales exceptions. A SKU showing positive stock and zero sales for a period where it normally sells is a strong phantom-stock signal. This check costs nothing and catches most cases.
Cycle count by value and velocity, not alphabetically. Count your fastest-moving, highest-margin lines monthly and your slow movers annually. Counting everything at equal frequency spends most of your effort on items where an error costs little.
Reconcile after every negative adjustment. A correction that takes recorded stock to zero usually means the outage began earlier than the record shows. Backdate it when estimating unfilled demand, or you will systematically under-count.
Stockouts you cause yourself belong in the same category. Allocating stock to one channel or location while another runs dry is recorded as fulfilled demand at the group level and as a stockout nowhere. If you operate multiple channels or sites, measure availability per location — a healthy aggregate figure routinely hides one site that is empty for a fortnight.
Benchmarks to Sanity-Check Your Rate
Use these to check whether your measured out-of-stock rate is plausible, not as targets.
| Benchmark | Figure | Source |
|---|---|---|
| Global average retail stockout rate | ~8% | NetSuite |
| Advertised / promoted products | ~2× the normal rate | NetSuite |
| US food retail, 2024 | 9.5% (from 12.3% in 2023) | Opensend |
| Global direct lost sales | >$1.2 trillion/year | Mirakl |
| eCommerce products with ≥1 stockout/year | 51%, averaging 35 days | Opensend |
The promoted-products figure deserves attention. Doubling the stockout rate exactly when you have paid to drive demand is the most expensive failure mode in the table, and it is usually a planning problem rather than a supply one — the promotion was forecast as a normal week. That is demand amplification in action, which our bullwhip effect guide covers in detail.
What to Do With the Number
Rank SKUs by stockout cost, not by stockout frequency. The item that runs out most often is rarely the item that costs you most. High-margin, high-substitutability lines belong at the top of the list even if they stock out rarely.
Set service levels per product, not per business. A blanket 98% target over-invests in staples customers will wait for and under-invests in discretionary lines they will not.
Stop short of 100%. Safety stock rises steeply and non-linearly as service level approaches 100%, so the final few percentage points cost far more than the sales they protect. The right target is where the marginal cost of another unit of safety stock equals the marginal cost of another stockout — for most businesses, the low-to-mid nineties.
Then decide whether forecasting is the fix. If your stockouts cluster around promotions, the fix is the promotional plan. If they cluster around long-lead-time items, the fix is reorder points. Only if they are spread across ordinary demand variability is better forecasting the right answer. Our predictive analytics for inventory management guide covers reorder points and safety stock, and our AI demand planning guide covers the forecasting side.
Conclusion
The reason stockout cost stays unmeasured is that the honest calculation is uncomfortable: it requires estimating demand you never saw and behaviour you never observed. So businesses either skip it, or accept a vendor's inflated version that multiplies every missed unit by full retail price.
The method above produces a defensible range instead of a persuasive number, and a per-SKU ranking instead of a headline total. The ranking is what you act on — it tells you which products deserve safety stock, and which ones customers will happily wait for.
Schedule a Strategy Call with Cogniq AI and we will run this against your own stock and sales history, including when the honest answer is that your stockout cost does not justify a forecasting project.
