A stockout alert is useful when someone can act on it. Detecting an empty shelf, a low inventory balance or a delayed delivery creates different tasks. Combining them into one red badge can send staff to replenish stock that is unavailable, or trigger a purchase when the product is already in the back room.
Begin with the operational distinction between shelf availability and recorded stock. A shelf observation is evidence about a visible location at a particular time. The inventory system represents quantities in business states and locations. Neither should automatically replace the other without an agreed reconciliation process.
Name the quantity and location
Shopify's inventory documentation distinguishes inventory states such as available and committed quantities. The exact states in your system may differ, but an alert must identify which quantity it uses and what that quantity means.
Record the store, product, shelf location and observation time. For camera-derived signals, retain the confidence or reason for uncertainty without assuming the camera knows the exact warehouse balance. If product placement changes, update the shelf mapping before interpreting the next detection as a stockout.
Worked example: stock exists but the shelf is empty
Suppose a hypothetical grocery store has twelve units of a product recorded in its back room. A shelf observation shows no visible units at 11:05. The system creates a replenishment task for the responsible aisle team, showing the expected shelf location and the current back-room balance.
The staff member checks the shelf and confirms the problem. If the back-room quantity is correct, they replenish and record the movement or completion evidence. If the product cannot be found, they open a stock-discrepancy task instead of marking the alert resolved.
Now imagine the camera was blocked by a trolley. Staff reject the observation with that reason, and the system waits for a fresh view. This feedback should improve alert review; it should not automatically relabel every future obstruction as a real stockout.
Deduplicate and prioritize tasks
A repeated observation should update the existing open task rather than create another task every minute. Use a stable key such as store, shelf and product, together with the incident lifecycle. A later recurrence after confirmed resolution can create a new incident.
Set priorities from agreed business rules: active promotion, expected demand or time waiting. Label uncertain signals separately. Staff should be able to see why a task is urgent and which source produced it. Avoid escalating an alert indefinitely when its underlying feed has stopped updating.
Preserve movement evidence
An alert is not an inventory adjustment. If resolution changes recorded stock, require the normal authorized adjustment workflow. GS1's EPCIS standard provides a framework for describing supply-chain events; its relevance here is the distinction between event evidence and an unexplained balance overwrite.
Keep timestamps and source identifiers for observations, replenishment actions and corrections. Integration retries must not create duplicate stock movements. A task can be closed with an operational reason while the related inventory discrepancy remains open for separate investigation.
Measure whether the workflow helps
Track verified alerts, rejected alerts, time to acknowledgment and time to useful resolution. Break results down by store and signal source. A high number of generated alerts is not evidence of improved availability.
Pilot on a limited product range and review staff workload. If most alerts lack a feasible action, change the detection rule or task design before expanding. A successful workflow connects trustworthy signals to accountable store actions and makes unresolved uncertainty visible.
Primary references
Related reading
Connect store signals and inventory actions through our business systems team.