Your production figures are written on a sheet next to the machine

Quantities, cycle times, stoppages: the data exists on the shop floor, gets copied twice, and reaches the office a day late.

Thirty minutes, no commitment

Describe your flow. I'll tell you what can be automated and where I would start.

An operator notes quantities and stoppages on a sheet. At the end of the shift the sheet goes up to the office. Someone copies it into a file. The following Monday, an approximate efficiency figure comes out of it that nobody really uses to decide anything, because it arrives too late and the numbers aren't trusted.

Let me be clear about what I will not sell you: not a replacement for your machine park, not a subscription analytics platform, and not predictive maintenance on equipment that lacks the sensors to make it possible. What this is about is far simpler — and it is what is actually missing.

What it looks like in your shop

The stoppage log exists, but the causes are three words

« Breakdown », « waiting », « setup ». Those three words cover very different situations, some of them easy to eliminate. Without distinguishing a material shortage from a tool change and from waiting for inspection, no targeted action is possible and you end up launching a generic improvement plan.

Counters are read once a day

The reading gives a total quantity, not a rate. So you cannot see the two-hour slowdown mid-shift, nor correlate a drift with a material batch or a tool change. The data is there but its resolution is too coarse to be useful.

The stated cycle time has not been checked in years

It comes from machine commissioning or from an estimate. The gap with the real time is paid for twice: in quotations and in scheduling, both of which rest on it.

Newer machines produce data nobody reads

The CNC control, the PLC or the press already log counters, alarms and running times. That information stays inside the machine because getting it out needs a skill the company doesn't have in-house. You are buying monitoring that already exists.

OEE is calculated for the meeting, not to act on

It is requested by management or by a customer, produced once a month with effort, and leads to no decision. An indicator that never changes a behaviour costs more than it returns.

What it actually costs

The point is not to measure for the sake of measuring. Here is where the absence of machine data genuinely costs.

Copying out readings: 30 to 60 minutes per dayOperator time at end of shift plus office time. Real, but not the main item.
Invisible micro-stoppages: 5 to 15% of running timeStoppages under five minutes are never written down — they are too short to justify writing. Added together they often exceed the long breakdowns, which do mobilise everybody.
Theoretical versus real cycle time: 10 to 30%That gap propagates through every quotation you issue. It is the direct link between this project and your margin, and it is the argument that convinces management.
Investment decided without measurement: a machine bought for nothingIt happens that a machine is bought for a saturation that actually came from changeovers or from waiting for inspection. Measuring first costs a fraction of the machine.

One point of honesty: this is the project with the least immediate return. It does not save administrative hours, it supplies the data that makes the other decisions correct. If your priority is a visible gain in three weeks, start with orders or supplier invoices and keep this one for later.

How it gets fixed

We work with the machine park you have, including machines from the 1990s, and without touching the PLCs that handle safety.

1. We look at what the machines already say
Many CNC controls and PLCs already expose counters and states, through a network port, a serial port or a standard protocol. One day on the floor, machine by machine, is enough to establish what can be retrieved without installing anything. It is free and it is almost always underestimated.
2. For silent machines, a simple sensor is enough
A sensor on the cycle lamp, a pulse counter, a dry contact: a few tens of euros and an intervention that modifies neither the machine nor its safety. You get run/stop and counting — which covers most of the need. I do not touch the safety side of equipment and I do not modify a PLC program in production.
3. The stoppage cause is entered by the operator, in one action
The machine can say it has stopped; it will never say why. A screen alongside offers six to eight causes defined with the team leaders, selected with one press. Six useful causes beat thirty boxes never filled in, and that list is the most important part of the project.
4. The data feeds what actually decides
Real cycle times fed back into routings and estimating, machine utilisation for the load view, top five stoppage causes displayed in the shop and reviewed weekly. Data that feeds no decision gets removed: that is a rule I apply to myself at the end of a project.

None of this requires connecting your shop to the internet. The network stays with you.

What this does not fix

What this does not fix:

This is not predictive maintenance. Predicting a failure needs vibration or current sensors, years of history and repeated failures. Across a varied SME machine park that is rarely fundable, and I do not sell that promise.
It does not replace basic preventive maintenance, which remains the highest-return activity in most workshops.
I do not work on machine safety functions, nor on the PLCs that provide them. That is a firm limit: equipment marking and conformity are not up for discussion.
It does not increase output. It tells you where output is being lost, and sometimes the answer is an organisation to change, not a machine.
On very old machines with no usable output and no way to fit a sensor, there is no sensible answer. I will tell you that rather than invent a fragile arrangement.

The questions I get asked

What does it cost?

The two-day audit is €1,500 and includes a survey of what can be retrieved from your machine park. A data-collection project across five to ten machines lands between €10,000 and €15,000 fixed price, excluding sensors and hardware, which you buy directly.

How long before it runs?

Three to four weeks for a first cell. Extending to the rest of the park happens machine by machine, without a new project.

Do our machines have to be connected to the internet?

No. Everything stays on your local network. It is a serious and frequent question: a shop connected without precautions is a risk, and many market offerings gloss over it. Here, the data does not leave your premises unless you explicitly decide otherwise.

Does this affect the warranty or CE marking of our machines?

No, within the scope described: reading existing information or adding a non-intrusive sensor on an indicator lamp. Any work touching safety functions is outside my scope and has to go through the machine builder.

Our machines are twenty years old — does it still work?

Often yes, and it is the most common case in SMEs. A dry contact or a lamp sensor works on a 1995 machine as well as on a new one. The feasibility survey during the audit gives you the answer machine by machine, with no commitment.

Is there a subscription?

No. No subscription, no per-machine licence. You pay for the project once and you own the result.

What if you're no longer available?

The code, the wiring diagrams and the documentation are with you. A local automation engineer can take over: it is built from standard, documented components, not exotic hardware.

In your trade

This job is not equally urgent everywhere. These pages say where it sits in the running order, trade by trade.

Let's talk for thirty minutes

You describe your flow, I tell you what can be automated, what cannot, and where I would start. If I'm not the right person, I'll say so during the call.