Lost sales

Lost sales are the revenue you didn't make because the stock wasn't there. When a customer wants to buy and the SKU is out, that demand doesn't disappear from the world — it disappears from your numbers. It's unmet demand during a stockout, and it's usually invisible.

That invisibility is the whole problem. Every other inventory cost shows up somewhere — holding cost on the balance sheet, markdown in the margin line, write-off in the P&L. Lost sales show up nowhere, because the transaction that would have recorded them never happened.

Why it's invisible

Your sales data records what did sell. During a stockout it records a zero — not the customers who landed, wanted to buy, and left. There's no line item for the order that wasn't placed. So the cost is real but unmeasured, and "unmeasured" tends to become "ignored."

It gets worse, because the forecast learns from that zero. A stockout suppresses your sales history: the period shows low or no sales not because demand was low but because you had nothing to sell. Feed that history into a forecast and it reads the gap as weak demand, lowers the next forecast, and you buy less — which causes the next stockout. The zero becomes self-fulfilling. A SKU can spiral down its own demand curve purely because each stockout teaches the model to expect less.

How to estimate it

You can't measure lost sales directly, but you can estimate them well enough to act on. The core estimate:

Expected demand during the stockout window × days out of stock × price.

Expected demand comes from the product's rate of sale before the gap, or from the forecast for that period — what the SKU would have sold per day had it been available. Multiply by the number of days it was out, then by the selling price, and you have a defensible figure for the revenue that didn't happen.

A worked version: a SKU selling 12 units a day at £40, out of stock for 9 days.

12 units/day × 9 days × £40 = £4,320 of lost sales

DTC brands can sanity-check this. During the gap, the product page kept getting traffic — multiply that traffic by the conversion rate the page ran at when it was in stock, and you get an independent estimate of the demand that hit a brick wall. If the two methods roughly agree, you can trust the number.

The estimate is deliberately conservative on one point: it assumes the per-day rate of sale would have held. For a SKU that was accelerating, or one that stocked out because it was selling fast, the true figure is higher.

Why it matters

Lost sales is the number that wins the safety stock argument. The case for holding more buffer always loses on its own, because the cost of carrying it — cash tied up, storage, markdown risk — is visible and easy to point at. The cost of not holding it is invisible until someone estimates it.

Put a pound figure on the stockout and the debate changes. "We'd hold £6,000 more stock" sits next to "we lost £4,320 in nine days last time this SKU ran out" — and now it's a comparison, not a one-sided cut. Quantifying lost sales is how availability stops losing the budget debate to visible holding costs by default.