You walk up to the line view: Filler 92% · Capper 61% red, downtime tile high, scrap normal · Labeler 89% · Case packer 91%.
Q1 What's your first move?
A
Check the labeler — 89% has room to improve too. A healthy station's two points are noise next to a red tile losing whole percentage points to stops.
B
Walk to the capper and find what's stopping it — the red OEE plus high downtime says stops are the cause. The tile names the station, its downtime neighbor names the cause. That's the read-then-walk discipline.
C
Note it for the end-of-shift report and keep your rounds. A live red tile is a now problem. Deferring it to a report is exactly what the live view exists to prevent.
Covered in Module 01
Q2 A station shows OEE 78% with downtime low but scrap high. Where is the loss coming from?
A
The OEE calculation must be wrong. The numbers are consistent: quality losses drag OEE even when uptime is perfect.
B
It keeps stopping — availability is the problem. Downtime is low; the station isn't stopping. Reading OEE alone would have sent you chasing the wrong loss.
C
It's running but making bad parts — a quality loss, not a stops problem. Low downtime + high scrap = the station runs but rejects parts. The neighbors tell you why the headline is low.
Covered in Module 01
Q3 Why does the line view rank as "one screen" instead of separate OEE, downtime, and scrap reports?
A
Because the low tile tells you where to look and its neighbors tell you why — the read only works side by side. OEE is the headline, downtime and scrap are the cause. Splitting them into reports breaks the glance-and-walk read.
B
To save screen space on the floor terminals. The layout is about the reading discipline, not pixels.
C
Because operators aren't given access to full reports. It's not about access — the one-screen view is faster than any report for catching a live loss.
Covered in Module 01
A jam stopped the line for six minutes. You cleared it and you're slammed catching up. The log asks for a reason.
Q4 What do you do?
A
Log it now with the specific jam code and the real down/up times. A true, timely entry puts the stop in the right bucket so the pattern surfaces and the AI ranks the real fix.
B
Skip it — six minutes won't move the shift's numbers. Small stops are exactly how micro-stop patterns hide. Skipped logs are invisible losses.
C
Tag it "Other" so the screen clears, and fix the entry later. "Other" hides the cause and "later" rarely comes — the pattern never surfaces.
Covered in Module 02
Q5 A 40-minute product changeover just finished. How does it get logged?
A
It doesn't need logging — it was planned. Planned doesn't mean free. Changeover overruns are one of the Six Big Losses, and unlogged ones can't be shortened.
B
As downtime, since the line wasn't producing. Mixing changeovers into downtime inflates breakdown numbers and hides both real problems.
C
As a changeover with its start and finish — not as downtime. Changeovers are planned loss with their own bucket and their own fixes (like faster setup), separate from breakdowns.
Covered in Module 02
Q6 The real cause of your stop isn't in the reason-code list. What's the clean move?
A
Use "Other" from now on for that failure mode. "Other" is where patterns go to disappear. The list should grow to match the floor's reality.
B
Pick the closest true code, and tell your lead so the missing code gets added. You keep the entry honest today and fix the list for tomorrow — instead of quietly training "Other" to swallow real causes.
C
Don't log it until the code exists. An unlogged stop is worse than an imperfect code — the time vanishes from the data entirely.
Covered in Module 02
Top card: micro-stops at the capper, $2,600/wk. While you're reading it, Station 2 jams and the line stops.
Q7 Order your next three actions.
A
Finish the capper card → then handle Station 2 → log both at end of shift. A stopped line outranks any card, and end-of-shift logging blurs both entries.
B
Clear the Station 2 stop → log it with its reason → return to the capper card and close it out. Live line-down first, log while it's fresh, then back to the ranked loss. That's the whole priority rule.
C
Clear Station 2 → skip the log since it's obvious → back to the card. The unlogged stop never joins its pattern — six minutes of invisible loss, every time it happens.
Covered in Module 03
Q8 The top card names a station and a loss, but you know the line and suspect a different cause. What is the card to you?
A
A starting point — walk, check, and log what you actually find, even if it contradicts the card. The card is ranked impact, not an order. Your close-out log is how the ranking learns the floor's reality.
B
An order — do exactly what it says regardless. You know the line; the AI knows the pattern. The system is built for your judgment plus its ranking, not obedience.
C
A suggestion you can ignore without logging anything. Ignore it silently and the same card returns every shift while the data drifts.
Covered in Module 03
An anomaly alert shows a temperature reading climbing out of its band. You walk over and your gauge check confirms the drift. Parts still pass.
Q9 What's the right response?
A
Make the smallest correction that brings it back in range, and log what you did. Confirm, correct small, log — the fix is a tweak now and a scrap bin later, and the log is handover gold.
B
Shut the station down until maintenance can look at it. Nothing has failed. A confirmed drift calls for the small correction the process allows, not a stoppage that costs more than the drift.
C
Watch it for the rest of the shift before touching anything. You already confirmed it. Watching a confirmed drift is just choosing the expensive version of the fix.
Covered in Module 04
Q10 Why is "the alert fired but parts are still passing" the best moment to act?
A
Because the gap between the alert and the reject limit is where the fix is still a small adjustment. That window is the entire point of anomaly detection — after it closes, the same drift costs scrap and a hold.
B
Because alerts auto-escalate to your supervisor if you don't respond. The reason to act is the economics of the drift window, not escalation mechanics.
C
It isn't — acting before failure is premature. Waiting for failure means the drift got free rein to turn into scrap. The alert exists to beat that moment.
Covered in Module 04
End of shift: you swapped a worn guide rail on Station 4 (running fine now), Station 6's reject rate is creeping up but still in range, and an AI card about the labeler is half-done.
Q11 Which handover leads with the right things?
A
"Labeler card half-done — see my log note. Station 6 rejects creeping, still in range — watch it. Swapped Station 4 guide rail, holding." Open work first, the trend second, the completed change last — each in plain words the next crew can act on.
B
Full shift totals for OEE, downtime, and scrap, then the three items. The numbers already carry forward. Burying open items under retyped totals is how they get missed.
C
"Swapped the Station 4 guide rail today. Everything else normal." Leads with the finished item and buries the open card and the creeping trend — the two things that need eyes.
Covered in Module 05
Q12 You didn't get to an AI next-best action this shift. The honest handover move is:
A
Name it and say why — out of time, waiting on parts, needs a supervisor call. Named with a reason, it gets picked up instead of dropped or duplicated by the next crew.
B
Say nothing — the card will still be in the queue tomorrow. Unnamed, it either gets dropped or the next crew re-does your half-finished work.
C
Mark it done so the queue starts clean for the next crew. A false "done" hides a live loss and teaches the ranking the wrong lesson — the worst of both worlds.
Covered in Module 05