Production traceability: tracking every batch and every part, for real.
A recall, a customer complaint, an audit asking where a lot came from: that's the moment traceability stops being paperwork and becomes how you defend production. Tracing means being able to answer in minutes — not days digging through binders and spreadsheets — three questions: where this part comes from, where the products made with that lot ended up, and who and what passed through the line, and when. This guide explains what production traceability really is, the difference between genealogy and progress, how the data is collected from the field, and when tracing is enough versus when you need to govern the line with an MES/SCADA.
Written byStefano Brachet Contol — CTO & Founder, IOMA
Published
8 min read
Updated
Updated
1 · Definition
What is production traceability, and why isn't it «just a log»?
«Traceability» is one word on the brochure; on the floor it's two distinct needs. The first is genealogy: reconstructing the history of a lot or a single part — raw materials, components, stations passed through, parameters, operators, timestamps. The second is retrievability: finding that history fast when it matters, i.e. when the complaint or the auditor arrives. A record nobody can query in time isn't traceability: it's archiving.
The difference carries legal and commercial weight. Supply-chain standards — ISO 22005 in food, IATF 16949 and OEM requirements in automotive, lot documentation in pharma and medical devices — don't ask «do you have the data?», they ask «how fast can you isolate and contain a suspect lot?». The good answer is measured in hours, not days. Traceability is what turns a recall into a manageable event instead of a warehouse-wide freeze.
Forward traceability
From the raw-material lot to finished goods: if the supplier flags a defect on lot X, which units, pallets and shipments contain components from it? This is what lets you contain a recall without freezing the whole warehouse.
Backward traceability
From the finished product to its origins: the part the customer is complaining about — which lots, machines, recipe and shift does it come from? This is what lets you find the cause and close the non-conformity.
Internal traceability
The path inside the plant: which stations the part crossed, with which process parameters, in what sequence. It's the part the ERP can't see and that only line data can tell.
Simple test: if answering «where did lot 2245 end up?» means opening a binder or cross-referencing three spreadsheets, you don't have traceability — you have archiving. Traceability is that same question solved with a search.
2 · Two different things
Genealogy and production progress: what do they solve, and why aren't they the same?
Genealogy and production progress are often confused because both «follow» production, but they answer opposite questions in time. Genealogy looks at the past: what was produced, with what, how. Progress looks at the present and the plan: how far along the order is, how many units are left, which station is down right now.
They serve different people. Genealogy is the tool of quality and management when something has to be investigated after the fact. Progress is the tool of the shift lead and production while the line runs, to see whether the schedule is being met. A traceability-only system covers the first very well; the second, once it becomes active control, moves into MES/SCADA territory.
They aren't two systems or two archives: they're the same recording read in two directions. That's why a traceability-only system covers genealogy well, while progress tracking — once it becomes active control — moves into MES/SCADA territory.
Genealogy (backward-looking)
The consolidated history of the lot or part: components, parameters collected, stations, timings, outcomes. It's the data that defends the company in an audit or a recall. It isn't there to decide now, it's there to reconstruct later.
Progress (real-time)
The state of the order as it runs: good units and scrap, pace, bottlenecks, stations active or in alarm. It's the data the shift lead needs to act within the shift, not the day after.
Where they meet
Recording progress in real time is what feeds genealogy: every completion, scrap and lot-change event becomes a row of the history. Track the present well and you get genealogy for free.
3 · On the field
How do you actually trace parts on the line? The data comes from the field.
Serious traceability isn't typed in by hand: it's read from the field. The data that matters — part completed, lot started, process parameter, scrap, recipe change — is born in the PLCs that run the machines. The technical job is to bring it up reliably, without slowing the machines, to a layer that records it and makes it searchable.
In practice the bridge between machine and traceability software is standard industrial protocols: OPC UA and, for Siemens equipment, native S7 access. The traceability software reads state and events from the PLC; the PLC stays the truth of the process and is not touched. It's a read, not a command: tracing is not controlling the line, and that distinction is what separates a traceability-only system from an MES/SCADA.
Data comes from the PLCs
Part counters, station states, process parameters, cycle events: they're already in the PLC. Traceability software reads them via OPC UA or S7 and stores them with timestamps, without manual data entry that introduces errors and delays.
Per lot or per part
Granularity depends on the process. In process/batch (food, chemicals) you trace per lot: quantities, recipe, batch parameters. In high-value discrete (automotive, electronics, devices) you trace per part, often with a serialized ID read by barcode or RFID at each station.
The field, not paper
Every point where an operator transcribes a number is a point where traceability breaks. The goal is for line data to enter on its own; manual input stays only where human judgement is needed — a quality outcome, a scrap note.
You don't need to replace your PLCs to trace. An OPC UA/S7 integration reads even from PLCs that have been in production for years: traceability is added on top of existing automation, it doesn't replace it.
4 · How much you need
Traceability alone or a full MES: when is tracing enough, and when do you need to govern the line?
This is where you decide how much system you actually need, and where the right money is spent. If the need is to observe and record — know what passed through the line, reconstruct genealogy, answer the audit — a dedicated traceability software is enough. If the need becomes governing the line — scheduling orders, managing recipes and format changes, showing live alarms and sending commands — then you're in MES/SCADA territory.
The difference isn't marketing, it's in kind. Traceability reads and records; it doesn't send commands to the PLCs and doesn't decide what the machine does. MES/SCADA supervises and controls: it receives orders, manages recipes, shows alarms in real time, talks to the ERP. Many companies start from traceability because that's the urgency — the audit, the recall — and add control later, reusing the same data and master data.
Traceability is enough → ARIA Track
When you need to know what passed through the line without governing it: ARIA Track is the traceability software that reads events from the field (OPC UA + S7, read-only) and reconstructs lot and part genealogy, with search by lot, exportable history and a live station monitor. It observes and records, it doesn't send commands to the PLCs.
When you need orders, recipes, live alarms and commands: ARIA MES/SCADA is the platform that supervises and controls, with the PLC staying the truth of the process. Track and MES/SCADA share data and master data, so the upgrade throws away nothing you've already collected.
Software is half the job. Bringing data up from the PLCs, integrating the ERP, securing the OT network: that's industrial integration, IOMA's craft. IOMA's MES service implements traceability and MES in your plant, with ARIA or on the system you already run.
Rule of thumb: if the question is «what happened?», traceability is enough. If it becomes «what should happen now?», you need MES/SCADA. Starting from the first and growing into the second, on the same data, is the cheapest path.
5 · In practice
Practical requirements: what you really need and how to start without over-sizing.
A traceability project almost always fails for the same reason: it starts too big. Before choosing software it pays to fix what you must be able to prove, at what granularity and to whom — the auditor, the customer, management — because that's what drives the cost, not the feature list.
The path that works is incremental: one line, the data that truly matters, a fast go-live, then extend. Traceability delivers value from the first lot recorded; it doesn't require digitizing the whole plant before it's useful.
Define the requirement, not the feature
Which standard or which customer is the request coming from? Forward, backward, or both? Per lot or per part? Within how many hours must you isolate a lot? This, not the feature list, sizes the project.
Map the data points on the line
Which PLCs, which stations, which events must be captured and at what granularity. This is where you learn what's already readable via OPC UA/S7 and what needs work on the automation.
Start with one line, then extend
A pilot line gives you real genealogy and real response times. With that data you make extension decisions — more lines, the upgrade to MES/SCADA — on numbers, not on vendor promises.
Account for the OT network
Reading from PLCs means putting production data on the network: segmentation and hardening aren't an extra, they're part of the project — all the more for anyone in scope for NIS2.
The software you start one line with: lot and part genealogy, search by lot, exportable history for the audit.
Read-only · OPC UA · S7
FAQ
Frequently asked questions about production traceability
Questions Plant Managers ask before starting an MES/SCADA project.
What is product genealogy?
It's the reconstructable history of a lot or a single part: which raw materials and components it comes from, which stations it crossed, with which process parameters, when and with what outcomes. It's what lets you answer, in minutes, the forward («where did lot X end up?») and backward («where does this part come from?») questions that arrive in an audit or a recall.
Do you need an MES to trace parts?
No. For traceability alone a dedicated software that reads events from the field and reconstructs genealogy — like ARIA Track — is enough, without the complexity of an MES/SCADA. You need an MES when, beyond tracing, you have to govern the line: schedule orders, manage recipes and format changes, show live alarms. Many companies start from traceability and add control later, reusing the same data.
How does it integrate with the PLC?
The traceability software reads from the PLCs via standard industrial protocols — OPC UA and native S7 access for Siemens equipment — collecting states, counters and events with timestamps. It's a read: the PLC stays the truth of the process and is not touched. You don't need to replace existing PLCs; the integration is added on top of automation already in production.
Better to trace per lot or per part?
It depends on the process and the requirement. In process/batch (food, chemicals) you usually trace per lot: quantities, recipe, batch parameters. In high-value discrete (automotive, electronics, medical devices) you trace per part, with a serialized ID read at each station. The criterion is practical: what is the smallest unit you must be able to isolate in a recall?
Where does the traceability data live?
Wherever makes sense for your plant: on-premise on the shop-floor server, or in the cloud if you'd rather not run the infrastructure. It's not the location that makes traceability, it's that the data is correct, available and searchable when needed. The choice is made on network constraints, latency and internal policy, not on a trend.
How much does it cost to start with traceability?
Less than a full MES, because the scope is narrower: you read and record, you don't control the line. The cost depends on how many PLCs and stations to integrate, on the granularity (per lot or per part) and on how much is already readable via OPC UA/S7. For pricing models and 2026 market ranges, with the public ARIA price list as a concrete reference, see the guide to MES and SCADA costs.
Go deeper
Related pages that quality and production typically open after this guide.
Describe the line and the problem on a 30-minute technical call: we tell you straight away whether it's in our wheelhouse. We typically reply within 24 hours. From the first call to commissioning, you speak to the same person.