Ivory & Deene is an Australian homewares retailer that picks and packs every order in-house. For years the shipping side ran on a label platform bolted onto Shopify. It printed labels. It printed them well. Nobody was angry at it.
The trouble was never the label. It was everything standing around it.
The setup that worked, right up until it didn't
Be fair to Starshipit: it is a mature, well-built product and it is genuinely good at the job it sets out to do. Rules, labels, carrier accounts, at volume. If a floor's bottleneck is "too many labels, too many manual steps", that is a strong pick and it stays a strong pick.
But a label platform is, by design, a layer. It takes an order that lives somewhere else, turns it into a consignment, and hands a tracking number back. Everything upstream of that is somebody else's problem: what the pickers walk, what the packer checks, what the stock number means when two stores sell the last one. On a small floor, that somebody is a person. Usually the same person, every morning.
Two systems, one order, and a human in the middle
Run Shopify and a label app side by side for long enough and you stop noticing the tax you pay. It shows up as small things that never quite go away:
- Two tabs open for one order, and a habit of checking both because neither is the whole truth.
- A pick list that came from a printer, a spreadsheet, or memory, because the label layer doesn't pick.
- Stock that agreed with itself right until a second store sold the same unit.
- A print dialog. Every time. Right printer, right size, click, wait.
- The end-of-day ritual of making sure Shopify knew what the warehouse had already done.
None of that is a crisis. It is a tax. You pay it in seconds, hundreds of times a day, and you only see the bill when you try to grow.
| On the floor | Shopify + a label layer | One system |
|---|---|---|
| Where the order lives | Two places, reconciled by a person | One queue, fed by Shopify in seconds |
| The pick | Printed list, or the label app's queue read backwards | Batched pick lists on an iPad, routed through the racks |
| Stock truth | Whichever system was updated last | One number, written back to every connected store |
| Printing | Print dialog, choose printer, choose size | Straight to the thermal printer, no dialog |
| Two orders, one box | Two labels, or a manual workaround | One parcel, one tracking number, both orders updated |
| Margin on an order | Revenue here, freight cost there, maths later | Landed cost carried through to the order |
Closer to Shopify than a label ever gets
Most shipping apps talk to Shopify. The difference is how much of the conversation they're actually in. Orders land seconds after checkout, not on a poll. Fulfilments, tracking numbers and status flow straight back, so a customer gets their tracking email because a parcel was packed, not because somebody remembered to sync.
Under that sit the unglamorous parts that decide whether you trust a system on a Monday:
- Every store you run, worked as one queue, with stock pooled across them or kept separate, whichever suits the business.
- A catalogue that heals itself. If a product changes in Shopify and a message goes missing, a background reconcile catches it. You don't find out from a picker holding the wrong item.
- Shopify's own fraud signals turned into an actual gate: risky orders are held before anyone picks them, and releasing one is a decision on the record.
- Digital and no-shipping items never reach the pick queue, so the floor only sees work that has a parcel attached.
The bench got quieter
The single biggest day-to-day change is the least impressive on paper: the pack bench runs on the keyboard, and labels come out without a print dialog. OrderOps Print sends each label straight to the right thermal printer at the right size. Hit print and it's already rolling.
Around that, the small mercies add up. Two orders going to the same address become one parcel with one tracking number, and both orders update. A morning's worth of labels prints as one stream instead of a click each. Barcode verification at the bench is there when you want a packer's honest second pair of eyes, and off when you don't. It ships turned off. A wrong label can be voided right up until the manifest closes, because after that the parcel belongs to the carrier and pretending otherwise helps nobody.
Picking stopped waiting on the wifi
Warehouse wifi dies behind steel racking. Every warehouse's does. Picking now runs on ordinary iPads that keep working through the dead spots and catch up the moment they're back in range, so nobody stands in an aisle watching a spinner. And because the system knows where stock physically sits, a batch comes out as a route through the racks rather than a list in order-number order. Same orders, fewer laps.
Stock stopped being an argument
This is the part a labels layer structurally cannot solve, because it never owned the number. With one system holding stock and writing it back to every connected store, the count on screen is the count on the shelf, and every unit carries its true landed cost from receipt through to the order it ships on. Margin becomes something you can read rather than reconstruct at the end of the month.
The carriers didn't change. Neither did the rates.
Worth saying plainly, because it is the first question any merchant asks about a migration: the carrier accounts came across as they were. Same contracts, same negotiated rates, billed by the carrier directly. Nothing pooled, nothing marked up. Australia Post, StarTrack, Team Global Express, TNT and DHL Express are live, with Aramex in beta, and quotes refresh at the bench as the box dimensions change.
What we'd tell another floor
If labels are the only thing slowing you down, a good label platform will fix that, and Starshipit is a good label platform. The question worth asking is a different one: how much of your day is spent keeping two systems agreeing with each other?
At Ivory & Deene the answer turned out to be more than anyone had counted. Picking, packing, stock and labels now sit in one place, the second tab is closed, and the floor is quieter for it. That is the whole pitch. There isn't a cleverer one.