Key takeaways
- Recording lots at goods-in and goods-out leaves the middle of the process undocumented.
- Traceability runs two directions: genealogy (backward) and impact (forward). Endpoint records answer neither well.
- Capture material identity, resource, process conditions and quality outcome at every operation.
- Test it by picking a finished unit at random and reconstructing its history.
In-process traceability means being able to follow material through every step of production — not just what arrived and what shipped, but what happened to it in between. Which lot went into which batch, on which machine, under whose operation, through which quality checks, and what was scrapped or reworked along the way.
Most manufacturers who believe they have traceability actually have endpoint traceability: lot numbers recorded at goods-in, serial or batch numbers recorded at goods-out, and an undocumented gap between the two. That works right up until somebody asks a question about the middle.
The two questions traceability has to answer
Traceability is not one capability. It is two, running in opposite directions.
Backward (genealogy): given this finished unit, what went into it? Which raw material lots, which machines, which operators, which process parameters, which inspection results. This is the question you get from a customer complaint or an audit.
Forward (impact): given this suspect raw material lot, what did it end up in, and where is that now? This is the question you get at 6pm when a supplier issues a recall notice — and the one that determines whether you quarantine 40 pallets or 4,000.
Traceability is only as good as its weakest operation. Endpoint records tell you the beginning and the end of a story whose middle is the part people actually ask about.
Why “we record lot numbers” is not enough
Three things break endpoint traceability in practice.
Lots get mixed
A hopper is topped up mid-run from a second lot. A partial pallet from last week gets consumed first. Two lots of the same component are open on the line simultaneously. Unless consumption is recorded at the point it happens, the link between input lot and output batch is an assumption.
The BOM is not what was used
Backflushing consumes materials against the theoretical bill of materials. It is fast and it is usually approximately right — but substitutions, overdoses and partial issues are precisely the events you most need traced, and they are exactly what backflushing erases. If your BOM management assumes the recipe was followed, your traceability inherits that assumption.
Scrap and rework disappear
A unit that fails inspection, gets reworked and then passes has a different history from one that passed first time. If rework is not captured as an event against the specific unit or batch, that history is lost — and with it the ability to spot that your field failures correlate with reworked stock.
Watch out for
The recall you cannot bound is the expensive one. If a raw lot was split across three production runs on two lines over a week, and the only linkage is a goods-in record and a shipping record, reconstructing the forward path means a manual investigation under time pressure — usually while a customer waits for an answer.
What you actually need to capture
In-process traceability requires linking four things at every operation, not just at the ends of the process:
- Material identity — which specific lot or container was consumed, recorded when it was consumed rather than inferred afterwards.
- Resource and time — which machine or work centre, which shift, which operator, and when.
- Process conditions — the parameters that matter for your product. For some plants that is temperature and pressure; for others, torque or cycle time. Read from the equipment where possible, because transcription is where accuracy dies.
- Quality outcome — the checks performed, their results, and the disposition of anything that failed.
Do that at each operation and genealogy becomes a query rather than an investigation. Skip it at any operation and you have introduced a blind spot that will be discovered at the worst possible moment.
The in-process inventory problem
There is a category of stock that endpoint systems handle badly: material that has left raw inventory but has not yet become finished goods. Work in progress sitting between operations, in a queue, in a dryer, on a rack awaiting inspection.
Your ERP typically sees this as a single WIP value. Physically it is dozens of identifiable, locatable, traceable things — and when there is a quality question, the WIP is often where the affected material actually is. Tracking in-process inventory as real, addressable stock rather than an accounting abstraction is one of the clearest dividing lines between a system that traces and one that reports.
Building it without instrumenting everything at once
Full traceability across every operation is the goal, not the starting point. A sequence that tends to work:
Start where the risk is. Identify the operations where a defect would be most expensive or hardest to contain. That is usually where materials combine, or the last point before the product becomes difficult to inspect.
Capture consumption at the point of use. Before adding process parameters or machine feeds, get real material issue recorded against real orders. This single change usually explains most of your unexplained material variance as a side effect.
Connect the machines that already talk. Most plants have equipment producing usable data that nobody is reading. Start there rather than with the retrofit that needs a capital request. Machine monitoring is often the cheapest accuracy win available.
Fold quality in, do not bolt it on. Inspection results that live in a separate system have to be manually correlated later — which means, in practice, that they will not be.
Try this
Run the backwards test. Pick a finished unit at random and reconstruct its genealogy from your current records: raw lots, machines, operators, process conditions, inspection results. Whatever you cannot answer is your next piece of work — and it is far better to find out now than when an auditor or a customer runs the test for you.
Where Capitán fits
Capitán MES carries end-to-end in-process traceability as a core capability — including in-process inventory, waste management and quality control alongside work orders, BOMs and machine data, rather than as separate modules you have to reconcile. Because it runs on a no-code platform, the traceability model can follow how your process actually works, and change when the process does.
If you are trying to work out where your current blind spots are, talk to our team — the backwards test above is a good place to start the conversation.
The short version
Recording what arrived and what shipped documents the ends of the process and nothing in between. Capture material, resource, conditions and quality at each step, and both the backward and forward questions become answerable in minutes rather than days.