Factories usually digitise in the order things are easiest to measure, which is close to the opposite of the order that helps.
Machines get instrumented because sensors are available. Dashboards get built because output is countable. Meanwhile the work order still travels on paper, material is issued from memory, and nobody can say with confidence which job consumed what.
The result is precise measurement of a process that is not under control. This article is about the order that works, and why the first step is the work order rather than the machine.
Start where the job is defined
The work order is the spine of a factory. It states what is being made, in what quantity, to which specification, for whom, and by when. Everything else — material, labour, machine time, quality, dispatch — attaches to it.
If the work order lives on paper, nothing else can be reliably connected. Material issued cannot be tied to a job. Output cannot be attributed. A quality problem found later cannot be traced to the batch that caused it.
Digitising the work order first sounds unambitious and it unlocks everything after it. The test of whether you have done it properly: can you take any finished item and say which order it belonged to, what was consumed making it, and who worked on it?
What the work order needs to carry
- What is being made, and the specification version
- Quantity ordered, and quantity actually produced
- Which stage it is currently at
- Material issued against it
- Who worked on it, and when
- Quality outcome, including rejects with a reason
Notice most of these are populated as the job progresses. That is the point — the work order becomes the record, not a document that describes an intention.
The sequence that works
1. Work orders. As above. The spine.
2. Material issue against the order. Not warehouse stock control in general — specifically, linking what left the store to the job that consumed it. This is where yield becomes measurable, and where the difference between standard and actual consumption stops being a guess.
3. Stage or operation tracking. Where each job currently is. In a multi-step process this single change usually eliminates the largest source of daily chaos — people walking the floor to find out what happened to something.
4. Quality outcomes with reasons. Rejects recorded against the order and the stage, with a cause. Quantity alone tells you there is a problem; the reason tells you whose it is.
5. Downtime attribution. Where the hours went, by cause. This is the highest-value measurement, and it comes fifth rather than first because it is far more useful once you can attribute it to specific jobs and stages. Production monitoring covers this in depth.
6. Machine instrumentation. Automated capture, once the categories are stable and people are using them.
Most factories attempt this list in reverse, starting at six.
Why material issue is the step that pays
If there is one place a small factory recovers real money, it is here.
Standard consumption says a job should use a given quantity. Actual consumption is what left the store. The variance between them is yield loss, scrap, theft, or an incorrect standard — and until issue is recorded against jobs, all four are invisible and indistinguishable.
Most operations we have seen discover their standards were set years ago and never revisited, which means every quotation since has been built on a number nobody has checked. That finding alone frequently changes pricing.
The paperwork trap
The most common failure in factory digitisation is asking operators to record more than they can while doing the work.
A system requiring five entries per job will get one, entered at the end of the shift, approximately. And approximate records are worse than none, because decisions get made on them with unwarranted confidence.
The design constraint is the same one that governs a busy counter: capture has to be a by-product of the work, not an additional task. In practice that means scanning rather than typing, defaults that are usually right, and the awkward outcomes — a partial completion, a reject, a stoppage — being as fast to record as the normal ones. That principle is in how Truffaire builds software.
If recording the truth is slower than recording the tidy version, you will get the tidy version.
Where this connects to everything else
A factory is not a closed system. The work order originates from a customer order or a forecast. Material comes from purchasing and stock. Finished goods go to a warehouse and then to a customer.
If production is tracked separately from stock and orders, the numbers will disagree — production says it consumed material that inventory still shows as held, or produced goods the warehouse has not received. That is the fragmentation described in what a business operating system actually is, and it is why we treat production as a set of movements against the same record rather than a separate system to reconcile.
What we have seen
Factory operations and production monitoring are among the ten systems we have delivered.
Two findings recur.
Stage visibility changes the day more than any metric. Before it exists, supervisors spend a meaningful part of every shift establishing where things are. After, that time returns. It is rarely the headline requirement and it is consistently the most appreciated change.
Standards are usually wrong. Once actual consumption is measured against standard for a few weeks, most factories find their standards no longer reflect reality — materials changed, processes changed, the standard did not. This is uncomfortable and valuable, because everything from quoting to purchasing depends on it.
Frequently asked questions
We are small. Is this overkill?
The sequence still applies; the tooling scales down. A small factory with digital work orders and material issue recorded against them knows more than a larger one with sensors and no attribution.
Do we need barcodes or tablets on the floor?
Something that makes capture fast, yes. The specific technology matters less than the number of actions per entry. Typing job numbers into a keyboard on a factory floor does not survive contact with a busy shift.
What about scheduling and planning?
Deliberately later in the sequence. Planning software built on unreliable consumption and duration data produces confident schedules that do not hold. Get the actuals first; then plan against real numbers.
How does this relate to ERP?
An ERP typically covers this and much more, at a cost and complexity that suits larger operations. For a small manufacturer the risk is a comprehensive system where only the forced modules get maintained. Start narrow, prove usage, extend.
Will operators resist?
They resist recording that takes time and gives nothing back. Showing the floor its own progress against target — rather than only reporting upward — changes that more reliably than any instruction.
Where to start
Take one product line. Digitise the work order for it, then record material issued against those orders for a month.
At the end of the month, compare actual consumption to standard. Whatever that variance is, it is money, and it was invisible before. That single exercise usually justifies the next step before any larger investment is committed.
SPEXA covers factory operations against the same record as stock, purchasing and orders. If you want a read on where to start in your operation, get in touch.