Connecting the shop floor to the ERP: production data for manufacturers in Thailand
How manufacturers get production data out of machines, paper and Excel and into the ERP - the staged approach that works without a seven-figure MES project.
Most manufacturers already own the two ends - machines that produce and an ERP that plans - and lose the truth in between: production written on paper, collected in Excel, typed into the ERP a shift or a week later, wrong by then. Closing the gap rarely requires a monolithic MES project; a staged approach - digital capture at the line, machine signals where available, a robust integration layer into the ERP - delivers most of the value in months. The result is one version of production truth: inventory that matches reality, costing based on actual consumption, and planning that reacts today instead of next Friday. Here is the staged route, from the assessment walk to the ERP.
What does the gap actually cost?
Walk any plant that runs on paper-and-Excel reporting and the same bill appears:
- Latency. Planning works on yesterday’s (or last week’s) production; expediting and firefighting fill the gap.
- Inventory drift. The ERP’s stock figures disagree with the warehouse because consumption is booked late and approximately; the safety stock that compensates is working capital doing nothing.
- Costing fiction. Standard costs built on assumed consumption and cycle times, uncorrected by actuals - so nobody truly knows which products make money.
- The retyping tax. Hours per shift writing down and re-entering what machines and operators already knew - the same tax office work pays everywhere, with a factory-sized rate.
- Quality archaeology. A customer complaint triggers a hunt through paper batch records instead of a query.
Why not just buy an MES?
Sometimes you should - a full Manufacturing Execution System is the right answer for complex, regulated or high-mix plants ready for that commitment. But an MES program is a large, long project, and the most common failure mode in mid-size plants is buying one to solve what is, at root, a data-capture and integration problem. The staged approach below delivers the majority of the value at a fraction of the cost and risk - and if a full MES comes later, every stage remains useful feedstock for it.
The staged route
Stage 0 - walk the flow. A gemba-style assessment of one product family: where is production data born, written, retyped, lost? The output is a map and a ranked list - gaps scored by hours, error cost and decision latency. (This assessment is small, fixed-price by nature, and an excellent first project with a new partner.)
Stage 1 - digital capture at the line. Replace paper where data is born: operator tablets or terminals for confirmations, counts, scrap and downtime reasons; barcode/QR scanning for material issues and transfers. No machine interfaces required - people and materials carry the data. Design rule: capture must be faster than paper for the operator, or it will lose to paper.
Stage 2 - machine signals where they pay. Newer equipment often speaks (OPC UA or vendor protocols); older machines can be retrofitted with simple sensors for run/stop and counts. Start with the bottleneck line - automatic downtime and count data there changes real decisions; wiring the whole plant on day one changes slideware.
Stage 3 - the integration layer into the ERP. The stage that makes the others pay: confirmations, consumption and movements flowing into the ERP promptly and reliably, with the engineering that any serious integration needs - logging, alerting, reconciliation, safe retries. This is also where an aging plant system gets an adapter instead of surgery, the same pattern as any legacy modernization.
Stage 4 - dashboards and the feedback loop. Once the data flows, the plant can see: live production boards, OEE-style views on the lines that matter, actual-vs-standard costing that finance finally trusts. Reports stop being Friday’s argument and become Monday’s plan.
The stages at a glance:
| Stage | What it delivers | Typical effort |
|---|---|---|
| 0 - Walk the flow | Map of data gaps, ranked by payback | Days |
| 1 - Capture at the line | Paper replaced where data is born | Weeks |
| 2 - Machine signals | Automatic counts/downtime on the lines that matter | Weeks, per line |
| 3 - Integration layer | Production truth flowing into the ERP, monitored | Weeks |
| 4 - Dashboards | Live visibility, actual-vs-standard costing | Weeks |
What does this look like in Thailand specifically?
Thailand is one of Southeast Asia’s manufacturing hubs, and the plants we see - Thai, Japanese, European-owned - share the same profile: solid ERPs (SAP, local systems, everything between), capable machines across three decades of vintage, and the gap between them bridged by paper and Excel heroes. The bill of materials for closing it is engineering labor: capture apps, adapters, integration plumbing, testing. That labor at Bangkok rates is what makes the staged route affordable to mid-size plants, not just corporate programs - and a local partner matters when Stage 0 is literally a walk on your floor.
In short
- The problem is the middle: production truth dies between the machines and the ERP - latency, drift, costing fiction, retyping.
- You usually don’t need an MES first: capture at the line → machine signals at the bottleneck → integration layer → dashboards, in months.
- Design rule for the floor: digital capture must beat paper for the operator, or paper wins.
- The integration layer earns the money - built with alerting, reconciliation and safe retries, not hope.
- Start with a walk-the-flow assessment: small, fixed-price, and it turns the whole topic into a ranked, priced list.
AD System Asia builds shop-floor capture, machine-data and ERP integration for manufacturers - our clients include Hitachi, Nidec, Covestro and Benetton Group companies. Invite us for the walk: the assessment is a small engagement with a map and a ranked list at the end.
Frequently asked questions
What does connecting the shop floor to the ERP actually mean?
Making production reality - what was produced, on which line, with what materials, scrap and downtime - flow into the ERP automatically and promptly, instead of being written on paper, collected into Excel and typed in at the end of the shift or the week.
Do we need a full MES to get production data into the ERP?
Not necessarily. A full Manufacturing Execution System is one answer; many plants get most of the value from a lighter staged approach: digital capture at the line (tablets, barcode scanning), machine signals where available, and a solid integration layer into the ERP - built in months, not years.
How do you capture data from old machines without network interfaces?
Around them: operator terminals or tablets at the line for confirmations and counts, barcode or QR scanning for material movements, and sensors retrofitted for basic signals like run/stop. Machines without any interface still produce data - through the people and materials that touch them.
What is the first step in a shop-floor integration project?
A gemba-style assessment: walk the actual flow of one product family, list every point where production data is written down or retyped, and rank the gaps by hours, error impact and decision latency. The first integration is then chosen by payback, not by fashion.
Why do production reports in the ERP disagree with the shop floor?
Because the data passed through hands and time: end-of-shift paper, Friday's Excel consolidation, corrections nobody traced. Every hop adds delay and drift. Automating capture and transfer removes the hops - which is precisely why finance and operations stop arguing once it's done.