What is the difference between an MES and an ERP?
The ERP plans and accounts: what to make, what it costs, what stock is worth. The MES executes and records: what is actually happening on the floor right now, on which machine, by whom, from which material lot. They need each other. Trying to run shop floor execution from an ERP usually fails because ERP transaction models are too coarse and too slow for station level work, and trying to run planning and costing from an MES fails for the mirror reason.
Do we need an MES if we already have an ERP?
Only if you need execution level detail the ERP cannot supply: real time work in progress, machine level performance, unit or lot genealogy, or operator level capture. If you are tracking production on paper and entering totals into the ERP at shift end, you have the data but not in time to act on it, and that gap is what an MES closes. If your current visibility is adequate for your decisions, we will say so.
Will operators actually use it?
That depends almost entirely on how many actions a routine task takes, and it is the question we design around first. If recording a completed unit takes one scan, it gets recorded. If it takes a form, a dropdown, and a confirmation, a paper tally appears within a week. We test station interfaces in real conditions with real operators before repeating the pattern, and we treat operator rejection as a design defect rather than a training problem.
Can you integrate our older machines?
Some of them, and we tell you which before you commit. Equipment varies from rich OPC UA data to a single output contact to nothing at all. We survey the asset base and produce a per machine statement of what is readable. Where a machine cannot be integrated, operator entry is designed in deliberately. Any vendor promising uniform connectivity across a mixed age asset base without surveying it has not looked at your plant.
How do you handle network outages on the floor?
Station clients buffer locally and continue working, showing clearly what is pending, then reconcile when the connection returns with defined conflict rules. This is treated as a normal operating condition rather than an error case, because in most factories it is. A system that stops accepting input when the network drops will be abandoned for paper the first time it happens during a busy shift.
What does full traceability actually require?
Capturing identity at the point of consumption, which means lot or serial numbers recorded as material is issued and as units progress, plus machine, operator, and process parameters bound to each unit or batch. The requirement is granularity at capture time. It cannot be reconstructed later from aggregate quantities, which is why the traceability decision has to be made before build rather than added afterwards.
How long does an MES implementation take?
A first line with work order execution, data capture, and ERP integration is typically a few months, with the floor survey and station design taking a meaningful share. Extending across lines is faster per line but machine integration on older equipment is the usual variable. A whole plant is a staged programme, and we would not recommend attempting it as a single cutover.
Can we start on one line rather than the whole plant?
Yes, and we recommend it strongly. One line proves the station design, the integration approach, and the reconciliation before the pattern is repeated at scale, and it gives operators on other lines a working example rather than a promise. Whole plant cutovers concentrate all the risk into one weekend, and MES risk is mostly in human acceptance rather than in software.
How does an MES relate to production planning?
The plan decides sequence and timing; the MES executes it and reports what actually happened, which then corrects the next plan. They are separate systems with a tight loop between them. Our production planning and scheduling page covers the planning side, including how actual capture from the floor makes the schedule realistic rather than aspirational.
What about OEE reporting?
OEE is computed from captured availability, performance, and quality rather than estimated, and its value lies in the downtime cause codes rather than the headline figure. A single percentage tells you little; knowing that most of the loss is one changeover on one machine tells you what to do on Monday. We design the cause code structure with the people who run the line, since a taxonomy they did not agree to will not be used honestly.
What do you need from us to start?
Access to the floor for observation, an equipment list with controller types and any existing connectivity, your ERP and its interfaces, your traceability obligations, and a representative product with its routing and bill of material. Time on the floor is the single most valuable thing you can give us, because it consistently reveals constraints no document mentions.
