Do we need machine sensors?

No. The view is built from the schedule, downtime log and receiving records you already keep. Machine data can be added later if the view shows it is worth it.

Who classifies the delay reasons?

The team, from a short list agreed with the plant manager. Free-text notes can be sorted into that list automatically, with a person checking the unsure ones.

No. It is a view over the records you have. If you later need real-time tracking or lot traceability, an MES may be the right answer.

Why production keeps slipping, in one view

Delivery dates slip and everyone has a theory. The reasons sit in three spreadsheets and a downtime log. Here is how they become one view the plant manager trusts.

The problem

Orders ship late often enough that sales has stopped promising dates. Every week someone builds a report by hand, and every week the meeting argues about whose fault it was.

What was really causing it

The facts exist but live apart: the schedule in one spreadsheet, machine downtime in another, material and paint delays in emails. Nobody can see an order's whole path from cut to weld to paint, so every delay gets explained by whoever speaks first.

The facts exist but live apart: the schedule in one spreadsheet, machine downtime in another, material and paint delays in emails.

What the research says

8.3% of planned production time lost to downtime in US discrete manufacturing (D. S. Thomas, NIST, Annual Report on the U.S. Manufacturing Economy: 2025 (NIST AMS 100-76))
$245B a year, the cost of that downtime (D. S. Thomas, NIST, Annual Report on the U.S. Manufacturing Economy: 2025 (NIST AMS 100-76))
$32B to $58.6B lost to defects a year, depending on how it is estimated (D. S. Thomas, NIST, Annual Report on the U.S. Manufacturing Economy: 2025 (NIST AMS 100-76))

Downtime and defects are both late-order causes, but a national average does not say which one is yours. That is the point of counting delay reasons in your own shop: the meeting stops arguing about anecdotes and starts with the biggest bar on the chart.

What we built

One view of every late order and why it was late, built from the records you already keep. Delay reasons get a fixed list, so they can be counted instead of argued about.

Collected Schedule, downtime and receiving records are pulled together every night.
Classified Each delay is given a reason from a short, agreed list.
Shown One page shows late orders, their reasons and the trend, by work centre and by week.
Alerted Orders at risk are flagged while there is still time to act.

What stays as it is

Your spreadsheets keep working while the view is built. The team keeps recording what it already records.

Where AI fits, and where it does not

Mostly none. This is data work: joining records and counting. AI can help sort free-text delay notes into the agreed reasons, with a person checking the ones it is unsure about.

What it is worth

What comes next

Scheduling changes based on what the view shows, and the same view for suppliers' delivery performance.

Questions people ask

Do we need machine sensors?: No. The view is built from the schedule, downtime log and receiving records you already keep. Machine data can be added later if the view shows it is worth it.
Who classifies the delay reasons?: The team, from a short list agreed with the plant manager. Free-text notes can be sorted into that list automatically, with a person checking the unsure ones.
Is this an MES?: No. It is a view over the records you have. If you later need real-time tracking or lot traceability, an MES may be the right answer.

Annual Report on the U.S. Manufacturing Economy: 2025 (NIST AMS 100-76), D. S. Thomas, NIST (2026)

Canonical URL: https://ugibo.com/case-studies/why-production-keeps-slipping