Shipped: Real-Time Ops Notifications on the Receiving Line
Real-time returns receiving notifications are live alerts that fire the moment something happens on the physical receiving line, a scan, a grade, a flag, a disposition, rather than waiting for a dashboard refresh or an end-of-day report. areturnz ships this as a standing feature now: every event on the line can push a notification to the people who need to act on it, within seconds of the event itself.
This is the second real changelog entry we are turning into a full write-up, following the scan-to-search receiving workflow post. That release rebuilt how operators find and process parcels on the floor. This one closes the loop by telling the right person, at the right moment, that something on that floor needs attention.
What actually shipped
The receiving line at NJ-01 already generates a dense stream of events: inbound scan, photo capture, AI grade assignment, confidence score, disposition decision, and any operator override. Before this release, most of that stream lived in the dashboard and in nightly exports. Someone had to go look for it. Now, the same events can trigger push notifications, in-app alerts, and webhook payloads in near real time.
Concretely, that means a returns manager watching for a specific SKU, a partner tracking their own tenant's volume, or a QA lead auditing low-confidence grades no longer has to poll the dashboard. The system tells them.
Event types covered at launch
- Inbound scan confirmed (parcel received at NJ-01)
- Grade assigned (A, B, C, or R, with the confidence score attached)
- Low-confidence flag raised for human review
- Disposition set (restock, liquidate, donate, destroy)
- Operator override logged against an AI grade
Each notification carries enough context to act on without opening a second screen: the return ID, the grade, the confidence score, and a link straight into the evidence bundle for that unit.
Why this matters for the receiving line
Speed on paper does not help if the humans downstream of it are still working off a stale view. Our median cycle from inbound scan to disposition sits at about 48 hours across the network. That number depends on people reacting quickly to what the line is telling them, not just on the AI grading fast. A flagged item that sits unseen in a queue for six hours because nobody got pinged adds six hours straight onto that cycle.
Real-time notifications also matter for accuracy oversight. Our AI-vs-operator match rate runs at about 99.6% across 180,000-plus processed returns, which means the exceptions are rare but real. Getting an operator's eyes on a low-confidence grade within minutes, instead of finding it in a batch report the next morning, is what keeps that number from drifting. See how the confidence math behind those flags actually works in confidence scores beat gut calls on the grading line.
Where partners feel it
Multi-tenant partners get scoped notifications tied to their own volume, respecting the same isolation used across per-tenant reporting and webhook delivery. A partner running their own returns desk on top of the network can now get an alert the moment one of their own parcels hits a low-confidence flag, without seeing anyone else's traffic.
Before and after: what changed on the line
| Workflow moment | Before real-time notifications | After real-time notifications |
|---|---|---|
| Inbound scan | Visible only in dashboard queue, checked manually | Pushed as an event the instant the scan completes |
| Low-confidence grade | Surfaced in next batch review pass | Flags an operator immediately for review |
| Disposition set | Found via report export or dashboard filter | Delivered as a notification with evidence link attached |
| Operator override | Logged silently, reviewed later in audit | Notifies QA in near real time alongside the audit log entry |
| Partner visibility | Delayed to daily or on-demand summary | Scoped, live alerts per tenant |
How it fits the rest of the system
Notifications do not replace the evidence bundle or the audit trail, they sit on top of them. Every alert links back to the same signed data available through the dashboard and the API: the outer label photo, the opened parcel, the item shot, the defect close-up, the grade, and the confidence score. Nothing about the underlying record changes, the difference is how fast a human finds out it exists. You can see what that underlying evidence actually looks like at the evidence sample page.
Delivery channels at launch include in-app alerts, email digests for lower-urgency events, and webhook push for partners who want to route notifications into their own tools. Configuration lives per user and per tenant, so a brand watching one SKU category is not drowned in noise from the rest of the network.
Frequently asked questions
What triggers a real-time notification on the receiving line?
Scans, grade assignments, low-confidence flags, disposition decisions, and operator overrides can all trigger a notification. Which events fire for which user is configurable per role and per tenant.
Does this replace the dashboard or the evidence bundle?
No. Notifications are an alert layer on top of the existing evidence bundle and audit trail. Every alert links back to the same photos, grades, and signed data already available through the dashboard and API.
Can partners get their own scoped notifications?
Yes. Multi-tenant partners receive alerts limited to their own volume, using the same isolation model applied to webhook delivery and per-tenant reporting.
Does this change the 48 hour median cycle time?
It supports it rather than resets it. Faster human response to flagged items and dispositions helps keep the existing 48 hour median from creeping up as volume grows.
Where can I see what shipped before this?
Check the product updates hub for the full run of shipped-feature notes, including photo evidence on every parcel and the scan-to-search receiving rework.
If you want to see real-time notifications running against your own return volume, contact areturnz and we will walk through it.
Related reading: Shipped: package lookup across the network
