On‑Shelf Availability (OSA)

OSA is the last link of the availability chain, and the out‑of‑stock rate on the shelf (OOS) is its complement. It is measured in three ways that answer slightly different questions: a manual audit counts gaps on the shelf at a given moment, POS‑based detection infers a gap from sales that stop against the expected rate, and shelf images from cameras or robots record the shelf state directly.

• OSA (audit) = SKU–store checks with the item on the shelf ÷ all SKU–store checks
• OSA (time‑based) = trading hours with the item available ÷ all trading hours
• OOS rate = 1 − OSA

On‑shelf availability in practice

The measurement method sets the number. An audit at store opening, after the night fill, comes out better than one at the evening peak, when the shelf has been emptied and not yet refilled, so two programs with the same definition but different audit times cannot be compared. POS‑based detection has a clear limit: it works for items that sell several times a day, where a few hours without a sale are a signal. For an item that sells a few units a week, a day without a sale is the normal state and the method cannot tell a gap from a quiet day – that is the territory of intermittent demand.

Availability also multiplies along the chain. If the distribution center delivers 98% of ordered lines, 97% of store stock records are correct and 97% of delivered goods reach the shelf on time, the shopper finds the item in about 92% of cases (0.98 × 0.97 × 0.97), although each link looks good on its own. A distribution‑centre service level therefore does not tell the store what the shopper sees. The gap between system availability and OSA points to store processes: backroom stock, shelf filling and phantom inventory.

Veritico STOCK accounts for stockouts when forecasting, so it predicts actual demand rather than sales cut short by empty stock, and its reports include stockouts (actual service level) and potential stockouts. What happens between the backroom and the shelf remains a store process that a forecast does not replace. See store replenishment, demand forecasting and the Albert case study.

Related terms

• Stockout
• Phantom Inventory
• Service Level
• Fill Rate