Glossary/Stockout
What is stockout?
A stockout is a SKU running out of sellable inventory while demand for it still exists. On a marketplace the listing either disappears from search or shows as unavailable, so the cost is not only the orders that did not happen: advertising that continues to run against an unavailable listing is spent for nothing, search ranking decays during the out-of-stock window and takes time to recover after restock, and a share of the shoppers who wanted the product buy a competitor’s version and do not come back. The visible lost revenue is usually the smallest of the four costs.
01/Formula
Formula
Direct lost contribution = daily demand × days out of stock × contribution per unit
Total stockout cost ≈ direct lost contribution
+ ad spend during the window
+ ranking recovery cost
+ permanently switched customersExample
A SKU sells 22 units/day at ฿180 contribution and runs out for 9 days during a campaign period. Direct lost contribution = 22 × 9 × 180 = ฿35,640 Ads still running on the listing ฿6,200 Estimated ranking recovery: ~2 weeks at reduced impressions The restock order that would have prevented it was worth ฿16,000.
02/In detail
What does a stockout actually cost?
Four things, and only the first is easy to count. Lost contribution is daily demand times days out times contribution per unit, and it is the number most sellers stop at. Advertising is the second: ads that keep running against a listing nobody can buy are pure waste, and on automated campaigns they often keep running because the campaign is set at the shop or category level. Ranking is the third and the most persistent, because marketplace search rewards recent sales velocity and a zeroed listing stops accumulating it; recovery after restock is measured in weeks, not days. The fourth is substitution — a share of buyers find a competitor and stay there — and it is the only one that never comes back.
Why do stockouts cluster around campaign days?
Because demand on a mega-campaign day can be several multiples of a normal day while replenishment lead times are unchanged, so the buffer that comfortably covers ordinary variation is consumed in hours. The compounding factor is that campaign demand is also the most valuable demand: the traffic is subsidised by the platform, the conversion rate is higher, and the ranking gained during a high-traffic window carries forward. A stockout on 11.11 therefore costs several times what the same stockout costs in an ordinary week, which is why campaign-aware cover — planning stock against the forecast for the specific day, not the trailing average — matters more than a higher buffer year-round.
Is a stockout always worse than overstock?
No, and treating it as such is how shops end up with dead capital. Both are errors in the same decision, and the correct trade-off depends on the ratio between the contribution lost per unit of missed demand and the cost of holding a unit that does not sell. For a high-contribution, non-perishable, stable-demand product the stockout cost dominates and a generous buffer is right. For a low-contribution, seasonal, or fashion-cycle product the overstock cost dominates and running thin is right. This is the newsvendor trade-off, and it is why a single service-level target applied to a whole catalogue is always wrong somewhere.
03/Why it matters
The trap, in one paragraph.
Stockouts cost more than the lost revenue, and the extra costs are the ones no report shows. Ads that ran during the window are pure waste, the ranking algorithm demotes listings that go unavailable during high-traffic periods, and on live commerce a stockout mid-session ends the velocity instantly. Because the costs are invisible and the restock cost is visible, shops systematically under-invest in cover.
Common mistake
Planning cover against average demand. Stockouts are caused by variance, not by the mean — a SKU selling 20 a day on average and 60 on a campaign day will run out even though the average is well covered. Cover has to be planned against the distribution of demand over the lead time, which is exactly what safety stock is for.
04/In DataGlass
How Stockout is used in DataGlass.
DataGlass forecasts demand per SKU and projects days of cover against current stock, incoming replenishment, and known campaign windows, then ranks restock candidates by the contribution at risk rather than by units. Listings that are advertised while low on stock are surfaced, since that combination is where the compounding cost sits.
05/Sources
- [1] Newsvendor model
The standard formulation of the stockout-versus-overstock trade-off, where the optimal service level is set by the ratio of understock to overstock cost.
- [2] Service level (inventory)
How target service levels translate into cover, and why a single catalogue-wide target misprices products with different contribution.
Keep reading