In reports, a stockout costs one sale. In reality, it starts a meter that keeps running long after the shelf is refilled. The customer who switches to a lower-margin substitute, the express freight ordered in panic, the hours spent arbitrating who gets served, the credit notes and disputes, and finally the share of customers who will not come back: every floor exists, none appears on the same accounting line, and most appear nowhere.
This is exactly what makes stockouts so poorly treated: their cost is real but scattered, while the cost of the inventory that would have prevented them is concentrated and visible. The trade-off is therefore made, structurally, against availability, without anyone ever having decided it in those terms.