J.P. Morgan's September global manufacturing PMI just hit 53.0 — its highest since February 2022, with factory production growing at its fastest pace since July 2021.

Most plant owners could tell you their output is up. Far fewer could tell you, off the top of their head, where the margin on that output went.

That gap is the whole story of this newsletter.

Data you have: production logs, ERP entries, spreadsheets, machine readouts, utility bills. It exists. It has existed for years.

Data you use: a number you check before a decision. A trend that changes what you do on Tuesday.

Most plants have the first kind in abundance and the second kind in almost none. The logs get filed. The spreadsheets get updated. The bills get paid. And when a line bleeds — scrap creeping up two points, a machine stopping for twelve unlogged minutes a day — nobody sees it, because nobody put the data where a decision could find it.

This is not a software problem. New dashboards sit on top of the same unread logs. The fix is dumber and harder: pick the five numbers that decide your margins, pull them from what you already record, and look at them every week.

Data you have is an archive. Data you use is a habit. The plants that make money know which one they're sitting on.

If this was useful, forward it to one person who runs a plant.

There's a kind of downtime every plant has and almost no plant measures: the 10-minute stop.

A jammed feeder. A sensor that needs a wipe. A line changeover that takes "just a few minutes." Nobody logs it. It's too small to be worth the paperwork, so it never makes it into a report. And because it never makes it into a report, everyone assumes it never happens.

But it happens. Four, five times a shift. Every shift. Every day.

Do the rough math in your head — you don't need a system for this. Five unlogged stops at ten minutes each is nearly an hour of lost production per shift. That's a full shift's worth of output disappearing every few days, and nobody's investigating because nobody sees it. The production numbers just come in a little lower than expected, and the team works a little harder to catch up, and everyone treats it as normal.

It's not normal. It's just invisible.

This is the difference between downtime you know about and downtime you have. The big stoppages get meetings. The small ones get shrugged off. But the small ones are the pattern — and patterns are where the money goes.

You don't fix this by buying a system. You fix it by deciding the small stops count. A tally sheet at the line. A thirty-second note per stop. Two weeks of that, and the pattern shows itself. Most plants are surprised by what they find. Some aren't — they suspected it all along and just never looked.

The data you already have can tell you more than you think. Start with the stops nobody writes down.

If this was useful, forward it to someone who runs a plant floor.

Every operation we've ever worked with had one cost that nobody could point to.

Not the big obvious ones — rent, payroll, raw materials. Everyone watches those. The invisible one is the cost that's spread so thin across the day that nobody feels responsible for it. The material that gets trimmed and swept up. The machine that stops for ten minutes and starts again before anyone logs it. The line that runs a little slower on night shift because nobody set a baseline.

Ask a plant manager what their waste costs and most will give you a number. Ask where that number came from and the room goes quiet. It came from an estimate. From a line item somebody filled in two years ago. It's not measured — it's assumed.

Here's the thing about assumed numbers: they're always polite. Real numbers usually aren't.

The uncomfortable part isn't finding waste. It's that the waste was there all along, visible in the data the operation already had. Nobody looked, because nobody was asked to look. The information existed in spreadsheets, in machine logs, in someone's memory of how the floor runs. It just never got put together in one place.

So here's the question, and we'd genuinely like answers: what's the most invisible cost in your operation?

Not the cost you complain about — the one you can't quite name. Reply to this post or forward it to the person who would know. We read every one.

Sometimes naming the cost is the hardest part of fixing it.

Offcuts, trimmings, startup scrap, rejected first pieces off a run — this material has a cost that rarely shows up anywhere a manager looks. The raw material was purchased. The labor to run it was paid. The energy to process it was spent. Then it went in a bin.

It stays invisible for a simple reason: nobody's job is to count it. Production reports count finished units. Quality reports count defects at final inspection. But the trim that falls off every unit, every shift, every day? That's just "the process." It never gets a line item, so it never gets a question.

Here's what to look at instead. Pull one week of production: units made, material purchased or issued. The gap between the two is your scrap. Then break it down by shift and by line. Almost always, you'll find one shift or one line generates noticeably more — and the difference between the best shift and the worst shift is pure margin, recoverable without buying anything new.

You don't need sensors for this. You need a sheet of paper at the end of each line and someone willing to write down what got thrown away for one week. Seven days of honest counting tells you more than a year of assumptions.

The waste was always there. The only thing missing was someone looking at it.

If this was useful, forward it to someone who runs a plant floor.