IoT for small factories: where to start

Most IoT projects in small factories fail not because the wrong sensor was chosen. They fail because the first thing installed was the interesting thing, not the expensive one.

Monitoring production room temperature is interesting. But if your biggest loss is a machine that stops for four hours before anyone notices, a temperature sensor will not solve that.

This article helps you choose what is worth monitoring first, with figures and a sensible order.

Start from the loss, not from the sensor

Before discussing technology, answer one question: in the last three months, which event was most expensive — that actually spent money or lost production?

The answer is usually one of these four, and each needs a different sensor.

• Machines stop without anyone knowing. You need vibration, current, or even just a stop button that reports. The loss is measured in production hours lost.
• Product is spoiled because conditions changed. You need temperature, humidity, or pressure. The loss is measured in wasted material.
• Material runs out without warning. You need tank level or weight measurement. The loss is production delayed waiting for material.
• Power use rises without explanation. You need per-section power measurement. The loss is visible directly on the bill.

Pick one. The most expensive one, not the easiest to install.

A realistic cost estimate

A small IoT project usually costs less than people imagine, and more than sensor vendors promise.

• Sensors and transmitters: roughly five hundred thousand to two million rupiah per point, depending on what is measured.
• Data gateway and network connection: two to five million rupiah, installed once per area.
• Dashboard and alerting: fifteen to forty million rupiah, depending on how much you want to see.
• Monthly running cost: data cards, power, and maintenance — usually under five hundred thousand rupiah for one small factory.

For three machine monitoring points with a simple dashboard, a reasonable initial budget sits around twenty-five to forty million rupiah. Compare that with the loss you calculated in the previous section: if one unnoticed machine stop costs ten million, the system pays for itself within three or four events.

The mistakes that most often sink IoT projects

• Installing too many sensors at once. Twenty points, six months, and the team trusts none of them. Start with three.
• Nobody is assigned to act on alerts. A system that sends alerts into a WhatsApp group with nobody tasked to respond makes people ignore every alert within two weeks.
• The network is not ready. Sensors need a stable connection in places that often have neither a socket nor signal. This is the most common problem in the field and the least often budgeted for.
• Data is collected but never read. If the dashboard is not opened weekly, the system is only gathering numbers.
• No maintenance. Sensors need cleaning and calibration. Without it, readings drift within a year and people stop trusting them.

The order that usually works

Pick the one machine that matters most. Install two or three measurement points. Run it for three months. Compare the loss before and after. Only then decide whether it is worth expanding.

This feels slower than installing everything at once, and it is slower. But in three months you have something far more valuable than a fully sensored factory: proof that monitoring actually reduces the loss.

If the first three months show no change, you have spent tens of millions — not hundreds — to learn that.

The next step

If you have one machine that often misbehaves, write down three things: how many times it stopped in the last three months, how long each event lasted, and what one hour of downtime costs you.

Those three figures are usually enough to decide whether monitoring is worth funding, and if it is, what needs measuring. We will reply with an estimated cost and installation time — including if the conclusion is that monitoring is not yet worthwhile for that machine.