MES Implementation Checklist: The First 30 Days on One Line
Most MES projects do not fail because the software is bad. They fail because the factory tried to digitize everything at once: six lines, four product families, every quality form, the maintenance program, and the ERP integration before the first operator ever logged in. The implementation that works is boring. One line, a handful of data objects, live in weeks. Here is the checklist we would run, in order.
Before Day 1: Pick the Line and the Loser Metric
Choose the line that hurts the most, not the line that is easiest to integrate. The pilot needs a reason to exist, and the reason is a number: late orders, scrap on changeover, audit prep days, or utilization nobody trusts. Write the number down. Thirty days later, that number is the entire ROI conversation.
Then pick the one station on that line where the pilot starts. A single CNC, a packing bench, an assembly station with a scanner. Everything below happens at that station first.
Week 1: The Six Data Objects You Cannot Skip
A manufacturing execution system is only as good as its master data. You do not need clean enterprise data to start; you need these six objects, and they can be typed in by hand on day one:
- Parts and materials: codes your people already say out loud. If the floor says "steel rod 40Cr", the system says "steel rod 40Cr", not "RM-40CR-040-REV-C".
- Lines and stations: the physical map of the pilot line, with a code per station that matches what is painted on the machine.
- Work orders: import tomorrow's schedule from the spreadsheet you already have. The MES is not replacing your planning process this month.
- Bill of materials and routings: for the pilot product only. One part number, its materials, and the station sequence.
- Users and roles: operators, a supervisor, and someone who can override. Pin codes or usernames, not email journeys; operators share terminals.
- One quality gate: a single checklist at completion, even if it is three questions. Quality can expand later; one gate proves the pattern.
Notice what is missing: no ERP integration, no machine sensors, no custom reports. Those are week 5 problems, and starting a MES-ERP conversation before the floor is live is how projects stall in a meeting room.
Week 2: Run Paper in Parallel, Deliberately
For a few days, the operators do their normal paper log and the terminal. This is not distrust; it is calibration. The paper log is ground truth, and the gaps you find ("the terminal says the run started at 08:00 but we started at 07:40 because the handover is earlier") are process decisions, not software bugs. Write each one down and decide. Every gap you resolve in week 2 is a data-quality argument you will not have in month 6.
This is also when you label things. A keyboard-wedge barcode scanner and labels on material bags and lot stickers change the terminal from "a screen to fill in" to "a scanner with a screen". Scan-heavy stations get adopted; typing stations get worked around.
Week 3: Machine Data, Only If It Is Already There
If the pilot machine already speaks Modbus or OPC UA to an existing HMI or gateway, connect it: counts and state feed OEE automatically. If it does not, start with operator-reported counts and states. A manual number that exists beats an automatic number that needs a controls project. Retrofit the sensors when the pilot has proven the workflow; see machine monitoring on existing machines for the escalation path.
Day 30: The Go/No-Go Questions
By day 30 the pilot station should answer yes to all of these:
- Do operators start, report, and complete work orders at the terminal without help?
- Does the supervisor trust yesterday's report more than the paper log?
- Is the loser metric measured, even if it has not moved yet?
- Can you add the second station yourself, in under an hour, without the vendor?
Four yeses: roll to the rest of the line, then the second line. Two or fewer: the problem is data objects or adoption, and scaling will multiply it. Fix the pilot; the rest of the flow follows.
Frequently Asked Questions
How long does a MES implementation take?
One station, live: days to weeks. One line with machine data: a month or two. A factory: as long as you let the scope grow before the first line proves itself. The 30-day pilot is the unit of progress; a full rollout is just many of them.
Do we need to clean up all our master data first?
No, that advice kills projects. Clean data for the pilot part only, and grow the clean set as you onboard products. Every factory's data is messier than its ERP suggests; a MES that requires perfection before go-live will never go live.
Should the MES integrate with ERP first?
Later. The first integration that matters is the operator's hands and the terminal. Import the schedule as a file or a paste; reconcile by export. Once the floor lives in the MES, the ERP handoff becomes a two-day project instead of a two-month one.
Run the Checklist on Your Line
Voltrus MES is built for this rollout shape: one line to start, stations added by the shop itself, scanner-first terminals, and every module included. See the live product.
See Voltrus MES