Case study 03 ยท Manufacturing

What can the factory actually make this week?

Canopies UK planned production by hand, from one spreadsheet and a wall of T-cards. GSD Lab built a working model of the factory floor that answers the question in minutes, and says plainly what is holding the week up.

The plan lived in a spreadsheet and in one person's head.

Every week someone sat down with an order export, a stock report and what they knew about the factory, and worked out by hand what should be made. That knowledge was real and hard-won, but it only existed in one place, it took hours, and nothing behind it could be checked.

Moulds made it harder. Several products share the same mould, so two orders can quietly collide, and the only thing stopping that was somebody remembering. The question the business kept asking, and could never answer quickly, was simple: if we take this order, when can we actually make it?

A model of the factory, not another to-do list.

So we built one. Every product, every stage, every mould and every person is in there, with the real minutes each job takes and who is allowed to do it. Off that model, the app answers the questions that used to need a morning: what has to be made, who is going to make it, and what is going to be late.

It started as a capacity calculator. It is now the thing the week is planned on, and each round of real use adds the next bit of how the factory truly works.

The weekly route

One export in. A week the floor can run, out.

01

The order export drops in

The weekly file from SAP is read straight in, and every line is matched to a product.

Human input
02

Stock comes off first

Anything already sitting in the warehouse is taken off before a single job is planned.

Automatic
03

The week gets planned

Each job lands on a named person and a day, within their skills, their hours and the moulds available.

Automatic, with a human override
04

The cards come out

Cards print for the board, and anything at risk is listed with the honest reason why.

Floor output
394 of 396Order lines matched automatically on the first real import
Hours to minutesThe weekly make list, built by the app instead of by hand
Every weekIn live use by the operations team since September 2026

What it took the guesswork out of

Honest answers, including the unwelcome ones.

01

Shared moulds can't clash

Two products that need the same mould can no longer be planned on top of each other without the app saying so.

Was tracked by memory
02

Nothing is planned twice

Work already covered by stock never reaches the floor, so the plan is what is genuinely left to make.

Was a manual cross-check
03

Late is explained, not just flagged

The app says whether an order is short of people or short of moulds, because only one of those is fixed by working late.

Was a judgement call
04

Overtime is real overtime

Extra hours are booked as the sessions the factory actually works, by name, not as a lump of spare minutes.

Was a rough estimate

Does your week depend on one person's spreadsheet?

If the plan only exists in somebody's head, it can be written down properly. Let's map how your work really flows and build from that.

Book a 20-minute chat
Book a chat