AI & Automation

Running a Factory From Another Country: What a Head Office Needs to See Every Day

November 11, 2026 · 6 min read · Etzal Group Team

A group headquartered in Singapore owns a plant in Batam that makes plastic components. Each Monday the plant sends a spreadsheet: last week's output, rejects and downtime. In one such report, head office learns that a moulding machine was down for three days the previous week, that an export order will ship late, and that the customer has already been told by the plant's own sales coordinator. Nobody hid anything. The reporting rhythm simply delivered the news six days after it could have been acted on. Factory visibility across a border is about shortening that delay without flying a manager over every week.

What head office really needs

Less than it thinks. Ask what decisions are made at head office.

  • Customer commitments: will orders ship on time?
  • Capacity: can we accept this new order?
  • Capital: which machine needs replacing?
  • Support: where does the plant need help that it cannot get locally?
  • Risk: is anything threatening safety, quality or a key customer?

Each needs a few numbers, daily or weekly, and the plant manager's reading of them.

The daily page

One page, by nine the next morning.

  • Output against plan, by line or product family.
  • Quality: reject rate, and any customer complaint received.
  • Downtime: total hours, the top three causes, and anything still down.
  • Orders at risk: those likely to miss their date, with the reason.
  • Materials: shortages expected in the next two weeks.
  • People: attendance, overtime, vacancies in key roles.
  • Safety: incidents and near misses.
  • A short comment from the plant manager: what happened and what is being done.

The comment matters as much as the numbers. Numbers without explanation invite wrong conclusions from far away.

Where the numbers come from

Start with what exists

  • Production counts from shift reports.
  • Downtime from the maintenance log.
  • Rejects from quality records.
  • Orders from the planning system or spreadsheet.

Agree on definitions first. Does output mean pieces made or good pieces packed? Does downtime include changeovers and waiting for material? Two plants reporting the same word with different meanings cannot be compared.

Then automate the weak points

Manual recording is weakest where it matters most: short stops go unrecorded, counts are rounded, reasons are guessed at the end of the shift.

  • Automatic counting on the bottleneck line, from the machine controller or a simple sensor.
  • Machine status: running, idle, stopped, from a signal already present on most machines.
  • A stop reason chosen by the operator on a tablet from a short list when a stop exceeds a few minutes.
  • Energy by line, where it is a major cost.

Then extend line by line. A full manufacturing execution system is not a precondition for knowing whether line 3 ran yesterday.

Make the connection reliable

  • Local first. Data is stored at the plant and synchronised, so a broken link loses nothing.
  • A separate, secured connection between plant systems and head office, not the office internet shared with everything else.
  • Production networks kept apart from office networks.
  • A local person responsible for the system, with remote support.

The same picture for everyone

A principle that prevents much resentment: the plant sees its data first, and sees everything head office sees.

  • Shift supervisors get the live view of their lines.
  • The plant manager gets the daily page before it goes anywhere.
  • Head office gets the same page with the manager's comment.

A system built only to watch the plant from abroad will be fed bad data. One that helps the supervisor run the shift gets accurate data as a side effect.

A review rhythm

  • Daily: the page is read. No meeting, unless a threshold is crossed.
  • Thresholds for a call: a line down more than a set number of hours, a customer order at risk, a safety incident, a quality escape to a customer.
  • Weekly, thirty minutes: trends, the top three losses, actions and their owners, help needed.
  • Monthly: costs, capacity, improvement projects, people.
  • Quarterly, in person: walk the floor. No dashboard replaces it.

Decision rights across the border

Write down who decides what.

  • The plant decides: daily scheduling, overtime within a budget, maintenance priorities, routine purchases, hiring of operators.
  • Head office decides: capital spending above a limit, new customers and pricing, changes to product specification, senior appointments.
  • Together: capacity commitments, large schedule changes, responses to major customer complaints.

Without this, visibility becomes interference: someone a thousand kilometres away asking why machine 7 stopped for twenty minutes.

Language, culture and time

  • Agree the working language of reports and keep terms consistent. A short glossary helps.
  • Expect bad news to travel slowly unless it is plainly welcomed. Thank the person who reports a problem early.
  • Mind public holidays in both countries when setting deadlines.
  • A one-hour time difference between Singapore and western Indonesia is small. Use it for a regular call at a fixed time.

Multi-plant comparison

With several plants, the same definitions allow fair comparison.

  • Compare trends, not only levels. Plants differ in product mix and equipment age.
  • Share what works. The best use of comparison is to move a good practice from one plant to another.
  • Avoid league tables that reward creative reporting.

Compliance and data

  • Employee data such as attendance and performance by named person is personal data. Send to head office only what is needed, in aggregate where possible, and follow the data protection rules of both countries for transfers.
  • Customer and product data may be covered by confidentiality terms.
  • Access is by role, logged, and removed when people change jobs.

Common mistakes

  • Reports that arrive a week late.
  • Fifty figures where eight would do.
  • No shared definitions.
  • Numbers without the plant manager's comment.
  • A system for head office that the plant does not use itself.
  • Buying a large system before knowing which numbers matter.
  • Manual stop reasons filled in at the end of the shift.
  • No thresholds, so everything or nothing triggers a call.
  • Unclear decision rights.
  • Punishing the bearer of bad news.
  • Never visiting.

We build production monitoring that serves the shop floor first and head office second, from machine signals to a daily page, as part of IoT and Control Systems. If your plant is in another country and the news arrives late, contact us.

Summary

A head office in another country needs one daily page: output against plan, quality, downtime and causes, orders at risk, shortages, people, safety and the plant manager's comment. Build it from existing records with agreed definitions, automate counting and machine status on the bottleneck first, and let the plant see its own data before anyone else. Set thresholds for calls, a weekly review, written decision rights and regular visits.

Frequently asked questions

What should a head office see from a remote factory each day?

Output against plan by line, quality rejects, downtime with its main causes, orders at risk of late delivery, safety incidents, and material or staffing shortages expected in the coming days. Six to ten figures on one page, available by the next morning, are more useful than a long report at the end of the week.

Do we need sensors on every machine to get this visibility?

No. Start with data that already exists: production counts, downtime logs, quality records and the order book. Add automatic counting and machine status on the bottleneck line first, where manual recording is least reliable and the value of accuracy is highest. Extend only where the first step has proved useful.

How do we avoid micromanaging the plant from a distance?

Agree in advance which decisions belong to the plant manager and which thresholds trigger a call from head office. Use the daily numbers to ask questions and offer help, not to issue instructions about individual machines. Review trends weekly together and keep the plant team as the first reader of its own data.

Author

Etzal Group Team

Etzal Group Team, helping businesses grow with technology. Contact us.

Related service

WhatsApp AI Chatbot

We build, train, and run a WhatsApp AI agent that answers customers 24/7, qualifies leads, and follows up.