Why in-house Shopify fulfilment breaks at ~100 orders a day

There's a volume where in-house fulfilment stops feeling clever and starts feeling fragile. For most Shopify brands packing their own orders it lands somewhere around 100 orders a day — earlier if your catalogue is wide, later if you sell three SKUs.

Below it, a spreadsheet, a printer and someone who knows where everything lives is genuinely a good system. Above it, the same setup quietly starts costing you money in ways that don't show up on one tidy line. Here's what actually breaks, in the order it usually does.

1. The pick stops being reliable

Memory-based picking has a ceiling. One person who knows the shelves can pick near-perfectly at low volume. Add a second picker, a seasonal temp, a similar-looking variant, a 4pm rush — and mis-picks creep in. Each one is expensive twice: the reship, and the customer who quietly doesn't come back.

The fix isn't "be more careful." It's scan verification — the system confirms the right unit is in the right box at the bench, so the mistake is caught before it's sealed, not by the customer.

2. You're re-keying orders between tabs

At low volume, copying an order from Shopify into a picking sheet is fine. At 100+ a day it becomes a part-time job that produces nothing — and every manual copy is a chance to fat-finger an address, a quantity, a SKU. Worse, the spreadsheet and Shopify drift, so nobody fully trusts either one.

The fix is one intake: orders land from Shopify automatically and move through the same pick → pack → ship path, so no human ever retypes an order to make it shippable.

3. Stock stops telling the truth

This is the quiet killer. "How many do we have?" has three different answers — on-hand, reserved (already promised to open orders) and available (what you can actually sell). A single number in a spreadsheet conflates them, so you oversell, or you sit on stock you think is spoken for. Across more than one shelf or room it gets worse fast.

The fix is real-time, location-level stock that keeps on-hand, reserved and available separate and honest — down to the bin.

4. There's no recovery path

The straight line — pick, pack, ship — almost never survives contact with a busy day. The bin is short two units. A barcode won't scan. A unit's damaged. In a manual setup, every one of these is a person walking off to find someone, and the order stalls. At volume, stalls compound into a backlog.

The fix is software that treats a short pick or a failed sync as a branch, not a dead end — it routes to a recovery step and keeps the rest of the orders moving.

Why hiring doesn't solve it

The instinct at this stage is to add people. But the four problems above aren't labour problems — they're information problems. More hands picking from memory means more mis-picks, not fewer; more people re-keying means more drift. Throughput goes up a little and error rate goes up with it. You end up paying more to ship worse.

What changes the curve is giving the people you already have a system that removes the guesswork: guided, scan-verified picking; one source of truth for orders and stock; and a process that bends when reality does.

Where we come in

That gap — past spreadsheets, below a £1,500-a-month enterprise WMS with a three-month implementation — is exactly what we built Bonsai WMS for: guided pick/pack/receive flows, real-time location-level stock, Shopify in and Sendcloud labels out.

We're currently hand-onboarding a small cohort of UK/EU Shopify brands as founding partners, and we set the whole thing up with you. If your operation is starting to feel fragile at the seams, that's exactly the kind of warehouse we want to build around.