The question "Where is my order?" often comes up because the store and the customer see different information. A manager copies the tracking number, checks the carrier by hand, and retells the status in a messenger. Delivery automation is useful when it links every shipment to its order and creates a clear action when something is delayed. The main result of such a process is being able to tell a normal delivery from one that needs intervention, quickly.
When a process needs automation
One order can consist of several parcels, especially when different suppliers ship the items. So a single "shipped" status is not enough. You need data on each part of the order and rules for how the overall result is derived. Define separately which data the carrier confirms and which the supplier reports. A created waybill does not yet prove the parcel was handed to the carrier.
Recommended workflow
01. Linking the shipment
Store the order number, parcel contents, carrier, and shipment ID. For split orders, track each parcel separately. The manager should see which items are already on the way and which are still waiting to ship.
02. Receiving events
Update the status through the carrier's available interface or a scheduled check. Keep the original status and the event time. A repeated event must not create a new notification or move the order back a stage.
03. A clear notification
Turn statuses into a short message with the order number and the action expected from the customer. Send it only on a meaningful change. Treat consent for service messages and the chosen channel separately from marketing mailings.
04. Exception queue
A delay, a wrong address, a refusal, or no movement creates a task for the person in charge. The task needs a response deadline and the shipment details. Once it is resolved, record the outcome so the exception never stays open forever.
Handling exceptions
Event order
Events can arrive late. Compare their times and the allowed status transitions. If the carrier sends contradictory data, keep both events and pass the case for review instead of making an automatic promise to the customer.
Notifications and limits
Limit repeat messages and check the contact before sending. A message delivery failure must land in the work queue. Do not declare an order fully delivered until every linked parcel is confirmed.
The first pilot
Launch the process for one carrier and one shipment type. Prepare tests for a partial shipment, a repeated event, a wrong tracking number, and a return to sender. First the manager approves the message texts. Then automate only the statuses whose source and meaning are unambiguous.
Checking the result
Compare the same process and a comparable workload before and after the pilot. Set target values based on your own starting point.
Delivery enquiries
The number of "where is it" requests per comparable number of shipments.
Response speed
The time between a problem event and the first action by the person in charge.
Status accuracy
The share of checked shipments with the correct state in the store's system.
Repeat messages
The number of unnecessary notifications caused by duplicate events.
Frequently asked questions
Can we promise an exact delivery date?
Show the source's confirmed estimate and what kind of estimate it is. If the carrier gives an approximate date, keep it as approximate. A date change must be visible to both the manager and the customer.
What if the supplier does not send a tracking number?
Create a separate "awaiting shipment" stage with a deadline and an owner. A reminder to the supplier helps, but a missing number must never be automatically replaced with an "in transit" status. Nexora Digital Systems helps design such a process around your systems and working rules. In the first conversation we can set the pilot's scope, the process owner, and how the result will be checked.
Sources
The documentation separates the work of fulfilling an order from the actual fulfilment, including partial shipments.
Sources checked on 04.10.2026. The implementation order and pilot in this article are Nexora's recommendations and need adapting to your systems. The examples are not results from client projects.
