Question 1

What makes an OEE number "inflated" rather than just low?

Accepted Answer

Height and honesty are different problems. A low OEE is a real number that happens to be small — that is a SOFT NUMBER, and it is fine to report as long as nothing is hidden. An inflated OEE looks healthy because time was taken out of the denominator: changeover minutes booked as "planned" beyond your own changeover budget, which drops them from planned production time. The grader computes your OEE both ways — as reported and as true, with those minutes put back — and the gap between them is the inflation. Any changeover moved out of planned time makes the line read INFLATED, whatever the true number is. Excluded micro-stops are reported too, but they do not move OEE (see below).

Question 2

How is the OEE actually calculated?

Accepted Answer

The world standard: OEE = Availability × Performance × Quality. Availability = run time ÷ planned production time, where run time is planned time minus downtime. Performance = (ideal cycle time × total units) ÷ run time. Quality = good units ÷ total units. Multiply them and run time cancels: OEE = good output at the ideal rate ÷ planned production time. That is why moving a minute between downtime and run time — excluding a micro-stop from downtime, say — shifts loss between Availability and Performance but cannot change OEE, while taking a minute out of planned time does. The "true" version puts over-budget changeover that was booked as planned back into planned time (and into downtime). A row whose numbers cannot all be true — more good units than total, or more output than the ideal cycle allows in the time the line ran — is refused rather than capped at 100%.

Question 3

Why does a little hidden changeover flip a genuinely good line to INFLATED?

Accepted Answer

Because an OEE that shrinks its own planned time is not an OEE — it is a different, more flattering metric wearing OEE's name, and once you accept it you can no longer trust any comparison, target, or capital decision built on it. The gate is deliberately dispositive and worsen-only: it can push a passing score down to INFLATED, and it can never lift a failing score up. The worked example ships a filler line at a true 74.3% — a genuinely strong number — that still reads INFLATED because 25 minutes of changeover were taken out of its planned time. Its cleaned-up twin, the same line with every stop logged and the changeover back in planned time, releases to TRUE OEE AS DESCRIBED at the same 74.3%. The point is not to punish the line; it is to make the reporting honest so the number means something.

Question 4

Is this predictive maintenance or a sensor/IoT tool?

Accepted Answer

No. It connects to nothing, reads no PLC, and needs no sensors, integration, or pilot. You enter shift facts you already have — planned time, logged downtime, units, good units, and the excluded micro-stop and reclassified changeover minutes — from your existing shift log or MES export. It is deterministic and offline: the same inputs always produce the same verdict, and there is no model, no probability, and no benchmark multiplier baked in. Every number on the page is computed from your own inputs, which is exactly what makes it defensible in a capital review.

Question 5

Does it score operators or shifts?

Accepted Answer

No. It scores an equipment number — a line or cell — never a person, an operator, or a crew. There is no field for who was running the line, and the verdict is about the integrity of the OEE figure, not anyone's performance. Use it to make the reporting honest, not to rank people.

Question 6

What do I actually get, and does the workbook match the tool?

Accepted Answer

You get a runnable Python engine (stdlib-only, no install beyond Python), a workbook that reproduces the identical verdicts (Start Here, Dashboard, OEE Honesty Grader tabs), a seven-line worked example, and two playbooks — an OEE-audit facilitator guide and an availability-integrity runbook. The engine, the workbook, and the on-page demo are verified to produce byte-exact matching numbers on the shipped sample, so whichever you run, the verdict is the same.

Your 85% OEE isprobably a story.

The most-quoted number on the plant floor is the one nobody checks.

Grade a line yourself. Watch a good number get caught.

How the gate works, in one image

One command reads your shift log and grades every line.

Three rules keep the number honest.

Reported vs. true, side by side

Planned time is dispositive

World-standard math, no multiplier

A reporting-integrity check, not a sensor or a scorecard for people.

Anyone who signs off on an OEE number or spends against one.

Build the honest number, then act on it.

Manufacturing & Distribution Ops Pack

CSV / Data-Export Schema & Quality Gate

The Supplier-Concentration Risk Gate

The honest answers.

Stop reporting a number
you can’t defend.