What breaks when you pass 100 orders a day
The failures are not random and they do not arrive together. Four things break in roughly the same order in most stores, and each is much cheaper to fix before it breaks than after.
1. Manual booking becomes the bottleneck
Reading an order in Shopify and typing it into the courier portal takes a minute or two. At twenty orders a day that is a chore; at a hundred and fifty it is somebody's entire evening, done tired, which is where wrong addresses come from — and a wrong address is an RTO you paid for twice.
The fix: book from the order itself, in bulk, so the consignment number lands back on the order automatically. That last part matters more than the time saved — it is what makes tracking and reconciliation possible later. See booking and CN printing.
2. The NDR queue stops being worked
At low volume, a failed delivery is a phone call someone remembers to make. At high volume it is a list nobody owns, and parcels age out of the reattempt window into returns by default.
This one is expensive precisely because it is invisible: nothing alerts you that a parcel became an RTO through inattention rather than through refusal. The fix is a single queue with the courier's reason attached, worked daily, with reattempt/hold/return as explicit decisions. See NDR.
3. Returns arrive faster than they are recorded
Boxes come back and pile up. Stock says they exist; the shelf disagrees, or vice versa. Then you sell something you do not have.
Worse, at this volume you can no longer tell which parcels the courier said it returned but never actually delivered back to you — and those are a real, claimable loss that quietly becomes a write-off.
The fix: scan returns in as they are unpacked, so arrival is an event with a timestamp rather than an assumption. See Return Scan.
4. Reconciliation gets skipped
The monthly courier reconciliation is the first thing to go when everyone is busy, because skipping it has no immediate consequence. The consequence is deferred and cumulative: short-pays stop being caught, claims expire unnoticed, and the profit figure drifts from reality with nothing signalling that it has.
The fix: match per parcel on shipment ID rather than by totals, and let deadlines drive the queue. See COD reconciliation and claims.
The pattern underneath all four
- Each failure is a manual process that scaled linearly with orders while the day did not.
- None of them announce themselves — they degrade quietly and are noticed a month later in the numbers.
- All four destroy the same thing: the link between a parcel and its money. Once that link is gone, no amount of analysis recovers it.
Which is why the useful preparation is not "get bigger systems" but "keep every parcel attached to its order and its money, from booking to settlement". Everything else follows from that.
See it on your own orders — install free