A stockout costs you far more than the sale you lost

When a size stocks out, the report shows one number: the units you could have sold and did not. It is a clean, comforting number, because it is bounded. You lost this many sales, at this price, so the cost was this much. Everyone nods and moves on to the next line.

That number is the smallest cost of the stockout, and treating it as the whole cost is why brands tolerate stockouts they should be furious about. The lost sale is the tip. Underneath it sit three costs that never make the report: the customer you lost, the margin you gave away forcing a substitution, and the damage you did to your own forecast. Add those up and a stockout costs multiples of the sale it took.

Availability is not a service-level nicety you optimize when there is time. It is the lever that decides whether a customer buys from you again, whether your next buy is any good, and whether your season clears at full price. It deserves to be planned at the level it actually breaks, which is size and door, not the tidy aggregate the category report shows.

Start with the customer. A shopper who wanted a medium and found the rail empty does not file a complaint. She buys the medium somewhere else, and if that somewhere else is good enough, she buys the next thing there too. The stockout did not cost you one sale. It cost you a customer relationship you spent real money to acquire, and the lifetime value that went with it. That cost never lands on the stockout line, but it lands.

Then the substitution. Some customers do not leave; they buy down. The customer who wanted the full-price style takes the marked-down alternative because it was there and roughly close. You booked a sale, so it looks fine, but you sold a lower-margin unit to a customer who was willing to pay full price, and you cleared markdown stock you might have sold anyway. You paid twice: once in margin, once in a full-price sale you converted into a discount sale with your own stockout.

Finally the forecast, which is the quietest and most expensive. Your demand model learns from sales history, and a stockout writes a lie into that history. The medium did not sell in week three because it was gone, but the data records zero sales, so next year's forecast reads the medium as weaker than it is and buys it shallower, which stocks it out again. A stockout is not a one-season event. It is a forecast poison that compounds until someone corrects for it.

Stack the three and the arithmetic is brutal. The lost sale is one unit of margin. The lost customer is the margin on every future purchase that shopper would have made, which for a brand with any repeat rate is many multiples of the single sale. The substitution converts a full-price sale into a discounted one and clears stock you could have sold anyway, so you lose margin on both legs of it. And the poisoned forecast quietly reorders the shortfall into next season. None of these three land on the stockout line, so the report shows you the cheapest cost and hides the expensive ones, which is exactly backwards from how you would want to see it.

You cannot post a 71 percent full-price sell-through while your hero sizes are stocking out in half your doors. The two facts are incompatible. Availability at the size and door level is not a downstream metric; it is upstream of nearly every number the business cares about.

Here is the trap. The category shows 92 percent availability and everyone relaxes. But that 92 percent is a blend, and blends hide their worst cases. The extra-large is sitting at 99 percent availability in every door, which drags the average up, while the medium, the size that carries the style, is stocked out in your top twenty doors. The number looks healthy and the business is bleeding, because the sizes that sell are exactly the sizes that stock out first.

This is why availability has to be measured at the level it breaks. A stockout is always a specific size, in a specific door, on a specific day. Roll it up to the category and you have averaged away the only view that would have let you fix it. The brands that hold availability do not have more inventory; they have a plan that watches the size-and-door grid instead of the category headline, and reallocates before the medium runs dry.

There is a timing point buried in this too. A stockout is not a moment; it is a duration, and the cost scales with how long the size stays gone. A medium that stocks out on a Friday and gets replenished Monday cost you a weekend. A medium that stocks out in week three and stays gone until the season ends cost you the whole back half of the style at full price. So the question is not only where availability breaks but how fast you catch it and close it. Watching the grid weekly and reordering monthly means every stockout runs for weeks before anyone acts. Watching it live and reallocating the same day means most stockouts are caught while they are still cheap, before the duration has a chance to run up the bill.

If the report only ever shows you the muted bar, you will keep making decisions that trade the violet one away. That is the whole problem with costing a stockout by lost sales alone: you optimize against the cheapest third and ignore the expensive two-thirds.

The fix has two halves. First, plan and hold availability at the size and door level, not the category. That means a size curve built per-style and per-door from demand, depth set so the sizes that sell hold ninety-plus percent availability all season, and in-season reallocation that catches a size drifting hot before it stocks out. You are not buying more; you are putting what you bought where the demand for that size actually is.

Second, stop letting stockouts corrupt the forecast. When a size was unavailable, the model should treat that period as censored demand, not as zero sales, so next year's buy reflects what the size would have sold rather than what it could not sell because it was gone. That one correction breaks the compounding loop that turns a single stockout into a recurring one.

Cost a stockout by the lost sale and you will tolerate it. Cost it by the customer, the substitution and the poisoned forecast, and you will plan availability like it is the number that runs the business, because it is.

The reframe is the whole point. As long as a stockout is a small, bounded lost-sale number on a report, it will always lose the argument for more inventory or faster reallocation, because the visible cost is too small to justify the effort. Price it correctly, with the customer, the substitution and the poisoned forecast included, and the calculus flips. Suddenly holding availability at the size and door level, catching drift the same day, and protecting the model from censored demand are not nice-to-haves. They are the cheapest insurance you can buy against a cost that was always three times bigger than the line item admitted.

Availability is the lever. Pull it at the size and door level, protect the forecast from its own blind spots, and the stockout stops being a line you shrug at and starts being the thing your plan is built to prevent.