Digitising a manufacturing operation: where to start

Ask a plant owner what a unit of their product costs to make. In most cases the answer is approximate: "roughly this much."

That is not a knowledge gap. Product cost is assembled from four sources — inventory, production, labour and energy — and each is tracked in a different system or ledger. Bringing them together by hand takes days, and by the time it is done the number is already out of date.

This article sets out where to start when digitising a manufacturing operation.

Five signs the time has come

• Product cost is unknown, or calculated once a month and late
• Stock figures are only reliable after a physical count
• You learn raw material has run out at the moment an order is placed
• Payroll depends on output but is calculated by hand
• Two departments quote different numbers for the same metric

Three of these mean a system is already needed. Five mean you are losing money every month without seeing it.

Which module comes first

This is where most implementations go wrong. Many start with CRM or sales because those are the most visible. In a plant the order is different.

1. Inventory. The foundation for everything: raw material receipts, issues, balances and stocktaking. Without this data, product cost cannot be calculated.

2. Production. Plans, norms, material consumed and output produced. This is where cost becomes a real number rather than an estimate.

3. Payroll. Linked to attendance and output. This block usually saves the most time.

4. Cash and settlements. Money movement and receivables.

5. Sales and customers. Only now does the system see real stock and know real cost when an order is taken.

6. Reporting and analytics. Once the five above are running, reporting produces itself.

What happens in the reverse order: a polished CRM runs, but the salesperson still phones the warehouse to check stock.

Why product costing comes out wrong

This is the most expensive problem in manufacturing. There are three causes.

Material is valued at an old price. Stock consists of several batches, each bought at a different price. If the system does not record which batch was consumed, cost becomes approximate.

Waste and rejects are excluded. If eight per cent of metal is lost in cutting and that never reaches the calculation, cost lands below reality. You may be selling at a loss without knowing it.

Indirect costs are not allocated. Energy, workshop rent, equipment depreciation. They stay in a general pool and are never spread across products.

The link between inventory and production

This is where the most data is lost in a plant.

How it should work: a production order is issued, material is written off against the norm, actual consumption is recorded, and the variance becomes visible. If the norm is 100 kg and 118 kg was actually used, that is a question worth asking. Equipment, material quality, or something else.

Seeing that variance is the single biggest benefit of the system. Eighteen per cent on one item looks trivial; across a month it becomes a serious figure.

Integrations specific to a plant

Weighbridge and camera integrations matter most: they take the data out of human hands. Disputes are then settled on the record rather than on memory.

How the rollout runs

Weeks 1-2. Understanding the process. Time on the shop floor, separate conversations with each department, collection of the forms in use. This stage usually reveals that part of the process is written down nowhere and lives in a few people's heads.

Weeks 3-4. Inventory module. The first block to go live, run in parallel with the old ledger for a month.

Weeks 5-8. Production and norms. The hardest stage, because norms are often undocumented and have to be established together.

Weeks 9-12. Payroll, cash, integrations.

Weeks 13-16. Reporting, training, full cutover.

Three mistakes

Starting without norms. The production module rests on norms. If they are undocumented, they have to be established first — that is a process engineer's job, not the system's.

Implementing everything at once. Change is absorbed slowly in a plant. Stage by stage, with a month of parallel running at each step.

Building only for the office. The shop floor terminal has to be different: large buttons, few fields, usable with gloves on.

What it costs

A ready system fits most plants — processes in metal forming, furniture, food and building materials are broadly similar. Building from scratch is justified when your process is genuinely distinctive.

In summary

Digitising a plant starts with inventory, not CRM.

A practical order:

1. Calculate the cost of three products two ways and look at the gap
2. Start with the inventory module and run it in parallel for a month
3. Document production norms — that is work before the system, not inside it
4. Add weighbridge and attendance integrations early; they take data out of human hands
5. Involve one or two shop floor supervisors from day one