MES projects usually go wrong before product selection, not during it. If nobody has decided what counts as one record, or what counts as output, the same problems appear whichever product you buy.
Choose what constitutes one record: a work order, a lot, or an individual serial.
Write down what counts as production quantity: how trial, scrap and rework are treated, whether operations double-count, and whether counting happens at issue or completion.
Count how many taps an operator needs to finish one operation.
If ERP exists, decide which side owns items, BOMs and orders, and which direction results flow.
Evaluation typically starts by comparing feature lists. What actually drags projects out is not a missing feature but an undecided rule.
The phrase "production quantity" alone needs agreement. Whether trial shots count, whether rework is counted twice, whether pre-inspection quantity is output — each changes the number. Without the rule, you come back to "the totals do not match ERP" after go-live.
Settle the four below and most later choices follow automatically.
Following the order matters more than perfecting each step.
Choose what constitutes one record: a work order, a lot, or an individual serial. This determines how often the floor has to enter data and how far you can trace. Serial level is powerful but heavy, so confirm what buyers or regulation actually require first.
Write down what counts as production quantity: how trial, scrap and rework are treated, whether operations double-count, and whether counting happens at issue or completion. This document later becomes the basis for the ERP interface.
Count how many taps an operator needs to finish one operation. Past three, it generally will not be followed. Strip out everything a machine signal can supply, and let people enter only the remainder.
If ERP exists, decide which side owns items, BOMs and orders, and which direction results flow. The moment both sides can edit the same data, they will diverge.
Patterns that recur in implementation discussions.
Simultaneous conversion carries the highest resistance and failure risk. Verifying one high-impact area and widening is faster in the end.
Adding fields "in case we need them later" degrades entry quality. Once inaccurate data accumulates, nobody trusts the numbers.
Starting with manual entry is fine, but which machines can supply which signals must be surveyed early. Finding out later can force a redesign.
Screens and entry flow have to be reviewed by the people who use them. First showing them at the demo makes changes far more expensive.
Confirm these are in place before starting.
The procedure is the same without any product. These shorten the steps.
Questions that come up during evaluation.
It varies, but the unit of management and counting rule can be agreed in days once the relevant departments sit down. What usually causes delay is that nobody has been given the decision.
Yes, though MES then owns master data such as items and BOMs, and that overhead has to be accounted for.
It is almost always an entry-flow design problem. If the operator gains nothing and only carries more burden, it will not hold. Include something the floor wants — next-operation visibility, drawing lookup — on the same screen.
It depends on scope and how many machines are connected. Starting with one area finishes quickly, and verifying before expanding often shortens the total elapsed time.
Terms and procedures on this page follow the public standards and official sources below. Figures are as of each source's publication date.
Reviewed by the IYULAB engineering team · Last updated:
Equipment mix and current workflow determine how the procedure is adapted.
Request a site survey